refactor(host/W6.2): extract the Linux zero-copy GPU plumbing into the pf-zerocopy leaf crate

linux/zerocopy/* (CUDA context/buffers + EGL/Vulkan dmabuf import + the isolated
import worker) and linux/dmabuf_fence.rs move wholesale into crates/pf-zerocopy,
so the coming pf-frame vocabulary crate (FramePayload::Cuda owns a DeviceBuffer)
and the pf-encode/pf-capture subsystem crates can reach the GPU plumbing without
the host orchestrator in between (plan §W6). Content stays Linux-only; the crate
compiles to an empty lib elsewhere, so dependents carry a plain dependency.

drm_fourcc deliberately does NOT move: it consumes the frame vocabulary
(PixelFormat), which sits ABOVE pf-zerocopy — it lives with capture for now and
moves into pf-frame next. cuda's ffi re-export bumps pub(crate)->pub (the raw
CUdeviceptr vocabulary is consumed across the crate boundary by the encode
backends). A crate::zerocopy shim module keeps every existing path valid until
capture/encode themselves move out.

Verified: Linux clippy -D warnings (pf-zerocopy --all-targets + host
nvenc,vulkan-encode,pyrowave --all-targets) + 17/17 pf-zerocopy tests + 321/321
host tests; Windows clippy nvenc,amf-qsv --all-targets Finished exit 0.

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

crates/punktfunk-host/src/linux/drm_sync.rs

@@ -4,7 +4,7 @@
 //! RETAINED BUT CURRENTLY UNUSED: producer-driven explicit sync is the "right" fix, but no
 //! compositor we target produces a usable sync_fd today — Mutter+NVIDIA fails buffer allocation
 //! (`error alloc buffers`, no cogl sync_fd), KWin/gamescope blit so they don't race at all. We sync
-//! zero-copy from the consumer side instead (see [`crate::dmabuf_fence`]). This module is kept,
+//! zero-copy from the consumer side instead (see `pf_zerocopy::dmabuf_fence`). This module is kept,
 //! verified (ioctl numbers + a live signal→wait round trip), ready to wire in the moment a producer
 //! gains working `SPA_META_SyncTimeline`.
 #![allow(dead_code)]