perf(encode/nvenc-linux): LN0 — sampled submit/retrieve-lock split, sub-frame caps, zeroReorderDelay

Latency plan §7 LN0 (Linux/NVIDIA encode follow-on):
- sampled PUNKTFUNK_PERF submit split (copy/blend/map/pic) in nvenc_cuda —
  the host loop's submit_us folds all four together; the D2D input copy is
  the LN2 zero-copy target and now measurable on its own
- sampled blocking lock_bitstream timing on pf-nvenc-out — in two-thread
  mode the host loop's wait_us wraps a non-blocking poll, so the real
  encode wait was measured by no timer
- caps probe + log SUPPORT_SUBFRAME_READBACK / SUPPORT_DYNAMIC_SLICE_MODE
  (LN1 sub-frame slice-output prerequisites, fleet visibility)
- explicit zeroReorderDelay=1 in the shared low-latency config (P-only +
  no lookahead has no reordering anyway; pins the bit against preset or
  driver drift; shared with the Windows backend)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

crates/pf-encode/src/enc/nvenc_core.rs

@@ -118,6 +118,9 @@ pub(super) unsafe fn apply_low_latency_config(cfg: &mut nv::NV_ENC_CONFIG, c: Lo
     cfg.gopLength = nv::NVENC_INFINITE_GOPLENGTH;
     cfg.frameIntervalP = 1;
     cfg.rcParams.rateControlMode = nv::NV_ENC_PARAMS_RC_MODE::NV_ENC_PARAMS_RC_CBR;
+    // Explicit zero reorder delay: with P-only + no lookahead there is no reordering to buffer,
+    // but pin the bit so no preset/driver default can ever slip a frame of reorder delay in.
+    cfg.rcParams.set_zeroReorderDelay(1);
     let bps = c.bitrate.min(u32::MAX as u64) as u32;
     cfg.rcParams.averageBitRate = bps;
     cfg.rcParams.maxBitRate = bps;