refactor(encode): drop the crate-wide allow(dead_code)

Inherited from the pre-extraction host crate root as scaffolding for backend
paths defined ahead of the build that used them. A census across every feature
combination on both platforms (flip it to `warn`, rebuild) found it was hiding
exactly two items — so it bought nothing while blinding the crate to future rot:

  - `vaapi::fourcc` — superseded by `pf_frame::drm_fourcc`; no call sites left.
  - `vulkan_video::open_opts` — test-only (the smoke tests use it to pass the
    RGB-direct request explicitly rather than through the env), now `#[cfg(test)]`.

Removing it surfaced a real latent defect the Linux census could not see:
`amf.rs`'s `percentile` / `drive_and_measure` helpers are used only by the
`#[cfg(feature = "amf-qsv")]` latency A/B benchmark, so a `--features nvenc,qsv`
build compiled the helpers with their caller gated out. They now carry the same
gate as their caller.

Verified `--all-targets -D warnings` on Linux (no features; shipped
nvenc+vulkan-encode+pyrowave) and Windows (no features; pyrowave; qsv; nvenc,qsv).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

crates/pf-encode/src/enc/linux/vaapi.rs

@@ -39,11 +39,6 @@ use super::libav::{
 };
 use ffmpeg::ffi; // = ffmpeg_sys_next
 
-/// `fourcc(a,b,c,d)` — DRM FourCC packing (`a | b<<8 | c<<16 | d<<24`).
-const fn fourcc(a: u8, b: u8, c: u8, d: u8) -> u32 {
-    (a as u32) | ((b as u32) << 8) | ((c as u32) << 16) | ((d as u32) << 24)
-}
-
 /// The render node a VAAPI/DRM device should open, from [`pf_gpu::linux_render_node`]: a
 /// matched web-console GPU preference pins it, else `PUNKTFUNK_RENDER_NODE`, else the single-GPU
 /// default.

crates/pf-encode/src/enc/linux/vulkan_video.rs

@@ -524,7 +524,9 @@ impl VulkanVideoEncoder {
     /// `open` with the RGB-direct request explicit instead of read from the env — the smoke
     /// tests use this (env mutation races parallel tests). `want_rgb` engages the RGB-direct
     /// source only if [`probe_rgb_direct`] also passes; otherwise the session opens on the CSC
-    /// path with the verdict logged.
+    /// path with the verdict logged. Test-only: the production entry point is [`Self::open`],
+    /// which resolves `want_rgb` from the env + the `cursor_blend` hint.
+    #[cfg(test)]
     pub(crate) fn open_opts(
         codec: Codec,
         width: u32,

crates/pf-encode/src/enc/windows/amf.rs

@@ -2760,6 +2760,10 @@ mod tests {
     }
 
     /// The `p`-quantile of `samples` (µs), sorting in place. `0` when empty.
+    /// Gated like its only caller, the `amf-qsv`-only §5.2 latency A/B below — otherwise a
+    /// `--features nvenc,qsv` build compiles this helper with the benchmark cfg'd out and trips
+    /// `dead_code` (which the crate root no longer blanket-allows).
+    #[cfg(feature = "amf-qsv")]
     fn percentile(samples: &mut [u128], p: f64) -> u128 {
         if samples.is_empty() {
             return 0;
@@ -2778,6 +2782,8 @@ mod tests {
     /// so its submit→AU is the bare ASIC time. The last ~2 unflushed frames on the ffmpeg path
     /// are left unmeasured (dropped with the encoder) so every recorded sample is a genuine paced
     /// submit→AU.
+    /// Gated like its only caller (see [`percentile`]).
+    #[cfg(feature = "amf-qsv")]
     #[allow(clippy::too_many_arguments)]
     fn drive_and_measure(
         enc: &mut dyn Encoder,

crates/pf-encode/src/lib.rs

@@ -5,9 +5,11 @@
 //! VA surface). One [`Encoder`] trait, selected in [`open_video`]. Extracted into a subsystem crate
 //! (plan §W6): depends on the shared frame vocabulary (`pf-frame`) + zero-copy plumbing
 //! (`pf-zerocopy`), never on capture — the capture→encode edge is one-way.
-// Scaffold: some backend paths + trait defaults are defined ahead of the per-feature build that
-// uses them (mirrors the host crate root's allow before the extraction).
-#![allow(dead_code)]
+// NOTE: no crate-wide `#![allow(dead_code)]`. It was inherited from the pre-extraction host crate
+// root as scaffolding for backend paths defined ahead of the build that used them, but a census
+// across every feature combination on both platforms found it was hiding exactly two items — so it
+// bought nothing and blinded the crate to future rot. Genuinely test-only helpers carry
+// `#[cfg(test)]` instead.
 // Every unsafe block in this module tree carries a `// SAFETY:` proof; enforce it (unsafe-proof
 // program). As a parent module this also covers the child modules (windows/linux backends).
 #![deny(clippy::undocumented_unsafe_blocks)]