feat(punktfunk/1): request-IDR recovery for a wedged client decode

Fixes the intermittent first-connect freeze. The host streams infinite GOP — one
opening IDR, then P-frames only (recovery keyframes just on loss) — so when the
client's decoder wedges on the cold first session (a lost/corrupt opening IDR, a
bad early P-frame) the picture stays frozen until the far-off next keyframe. The
client had no way to ask for one; now it does.

Add a RequestKeyframe control message (client -> host, reliable control stream),
mirroring Reconfigure:
- core: quic.rs RequestKeyframe (type 0x03) + roundtrip test; client.rs
  CtrlRequest::Keyframe + NativeClient::request_keyframe; abi.rs
  punktfunk_connection_request_keyframe (header regenerated).
- host: m3.rs decodes it in the control loop and signals the encode loop, which
  coalesces a burst and calls enc.request_keyframe() — wiring the existing
  NvencEncoder hook (force_kf -> next frame pict_type=I), the same recovery the
  GameStream path already had via force_idr.
- apple: PunktfunkConnection.requestKeyframe(); StreamPump (stage-1) requests on
  layer.status==.failed; Stage2Pipeline (stage-2) on a sync submit failure and on
  the async decode-error callback via a thread-safe KeyframeRecovery. All
  throttled to <=1/250ms (the decode stays wedged for several frames until the IDR
  lands, so per-frame requests would flood the control stream).

Self-healing: a lost recovery IDR is re-requested after the throttle; the host
coalesces bursts into a single IDR.

Validated: cargo fmt + clippy clean; core + host test suites green (incl. new
request_keyframe_roundtrip); swift build + test (39 passed); xcframework rebuilt
(all 5 slices), header regenerated with no unrelated drift.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

clients/apple/Sources/PunktfunkKit/PunktfunkConnection.swift

@@ -332,6 +332,21 @@ public final class PunktfunkConnection {
         _ = punktfunk_connection_request_mode(h, width, height, refreshHz)
     }
 
+    /// Ask the host's encoder to emit a fresh IDR keyframe now — recovery when the local
+    /// decoder has wedged. The host opens the infinite-GOP stream with one IDR and then sends
+    /// P-frames only, so a stalled decode (a lost/corrupt opening IDR, a bad early P-frame —
+    /// most likely on the cold first connect) would otherwise stay frozen until the next
+    /// loss-triggered recovery keyframe, which may be far off. Fire-and-forget; the recovered
+    /// keyframe is the only ack. THROTTLE at the call site — the decode stays wedged for
+    /// several frames until the IDR lands, so requesting every frame would flood the control
+    /// stream. Silently dropped after close.
+    public func requestKeyframe() {
+        abiLock.lock()
+        defer { abiLock.unlock() }
+        guard let h = handle, !closeRequested else { return }
+        _ = punktfunk_connection_request_keyframe(h)
+    }
+
     /// The currently active session mode (updated by accepted `requestMode` switches).
     public func currentMode() -> (width: UInt32, height: UInt32, refreshHz: UInt32) {
         abiLock.lock()

clients/apple/Sources/PunktfunkKit/Stage2Pipeline.swift

@@ -44,11 +44,36 @@ private final class PumpToken: @unchecked Sendable {
     func cancel() { lock.lock(); live = false; lock.unlock() }
 }
 
+/// Throttled host keyframe requests for decode recovery. The decoder's async error callback
+/// (a VT thread) and the pump thread (a submit failure) both signal a wedge; this coalesces
+/// them so the control stream isn't flooded while the decode stays stalled for several frames
+/// until the requested IDR lands. Bound to the live connection in `start`, unbound in `stop`.
+private final class KeyframeRecovery: @unchecked Sendable {
+    private let lock = NSLock()
+    private var connection: PunktfunkConnection?
+    private var lastNs: UInt64 = 0
+
+    func bind(_ c: PunktfunkConnection?) {
+        lock.lock(); connection = c; lastNs = 0; lock.unlock()
+    }
+
+    func request() {
+        lock.lock()
+        let now = DispatchTime.now().uptimeNanoseconds
+        let due = lastNs == 0 || now &- lastNs > 250_000_000 // ≥ 250 ms since the last request
+        if due { lastNs = now }
+        let conn = due ? connection : nil
+        lock.unlock()
+        conn?.requestKeyframe()
+    }
+}
+
 public final class Stage2Pipeline {
     private let ring = ReadyRing()
     private let presenter: MetalVideoPresenter
     private let decoder: VideoDecoder
     private let presentMeter: LatencyMeter
+    private let recovery = KeyframeRecovery()
     private var token = PumpToken()
     private var offsetNs: Int64 = 0
 
@@ -63,9 +88,13 @@ public final class Stage2Pipeline {
         self.presenter = presenter
         self.presentMeter = presentMeter
         let ring = ring
+        let recovery = recovery
         self.decoder = VideoDecoder(
             onDecoded: { ring.submit($0) },
-            onDecodeError: { _ in /* the pump resets the session via reset() on the next IDR */ })
+            // Async decode failure (a bad P-frame referencing a lost/corrupt IDR): the pump
+            // resets to re-gate on the next IDR, and we ask the host to send one now (infinite
+            // GOP — it wouldn't otherwise come soon). Throttled in KeyframeRecovery.
+            onDecodeError: { _ in recovery.request() })
     }
 
     /// Start pulling AUs into the decoder. `onFrame` fires per AU at receipt (capture→client
@@ -77,9 +106,11 @@ public final class Stage2Pipeline {
         onSessionEnd: (@Sendable () -> Void)?
     ) {
         offsetNs = connection.clockOffsetNs
+        recovery.bind(connection) // arm host-keyframe recovery for this session
         token = PumpToken() // fresh token per start — cancel is permanent (like StreamPump)
         let token = token
         let decoder = decoder
+        let recovery = recovery
         let thread = Thread {
             var format: CMVideoFormatDescription?
             while token.isLive {
@@ -92,8 +123,10 @@ public final class Stage2Pipeline {
                     guard let f = format, token.isLive else { continue }
                     if !decoder.decode(au: au, format: f) {
                         // Submit/decoder error: drop the session and re-gate on the next IDR's
-                        // in-band parameter sets (a delta frame can't recover) — stage-1's policy.
+                        // in-band parameter sets (a delta frame can't recover) — stage-1's policy
+                        // — and ask the host for that IDR now (infinite GOP; throttled).
                         decoder.reset()
+                        recovery.request()
                     }
                 } catch {
                     if token.isLive { onSessionEnd?() }
@@ -125,6 +158,7 @@ public final class Stage2Pipeline {
     public func stop() {
         token.cancel()
         decoder.reset()
+        recovery.bind(nil) // stop requesting keyframes once the session is torn down
     }
 
     deinit { token.cancel() }

clients/apple/Sources/PunktfunkKit/StreamPump.swift

@@ -41,6 +41,7 @@ final class StreamPump {
 
         let thread = Thread {
             var format: CMVideoFormatDescription?
+            var lastKeyframeRequest = Date.distantPast
             while token.isLive {
                 do {
                     guard let au = try connection.nextAU(timeoutMs: 100) else { continue }
@@ -49,13 +50,19 @@ final class StreamPump {
                         format = f // refreshed on every IDR (mode changes included)
                     }
                     if layer.status == .failed {
-                        // Decode wedged: flush and re-gate on the next in-band parameter
-                        // sets — resuming with a delta frame can't recover. (A
-                        // request-IDR channel on punktfunk/1 is a host-side TODO; with the
-                        // host's infinite GOP this may otherwise stay black until the
-                        // next recovery keyframe.)
+                        // Decode wedged: flush and re-gate on the next in-band parameter sets
+                        // (resuming with a delta frame can't recover), AND ask the host for a
+                        // fresh IDR. With the host's infinite GOP the next keyframe could be
+                        // far off, so without the request the picture stays frozen — the
+                        // intermittent first-connect freeze. Throttled: the layer stays .failed
+                        // across several polls until the IDR lands, and one request suffices.
                         layer.flush()
                         format = AnnexB.formatDescription(fromIDR: au.data)
+                        let now = Date()
+                        if now.timeIntervalSince(lastKeyframeRequest) > 0.25 {
+                            connection.requestKeyframe()
+                            lastKeyframeRequest = now
+                        }
                     }
                     guard let f = format,
                           let sample = AnnexB.sampleBuffer(au: au, format: f),