io/pipeline: bump WRITE_PIPELINE_DEPTH 16 → 32

The depth was conservative because pre-0.21.8 a full sync_file_range
stall on NFS could traverse this channel and pin the producer. With
0.21.8's restored Phase 2.5 writer thread + 128 MiB byte-bounded ring
inside WritebackFile, every blocking syscall happens downstream of
this channel — never on it. The original "smaller buffer reduces
backpressure risk" rationale no longer applies.

Empirical (2026-05-15 Civil War UHD remux on 0.21.8): 30.9 MB/s
sustained but instantaneous samples spanning 9-53 MB/s, stdev 9.3.
Distribution clusters 52% of samples in 25-35 MB/s but has a long
9-15 MB/s tail. The tail corresponds to brief stalls in the matroska
builder when the sink momentarily lags — exactly the case a deeper
inter-thread channel covers. 32 frames at PES-frame sizes is still
well under a megabyte of additional memory, so the cost is zero.

src/io/pipeline.rs

@@ -92,10 +92,16 @@ pub const DEFAULT_PIPELINE_DEPTH: usize = 4;
 /// consumer blocks on write.
 pub const READ_PIPELINE_DEPTH: usize = 32;
 
-/// Write pipeline depth. Smaller buffer reduces backpressure risk when
-/// sync_file_range blocks; prevents producer from accumulating too much
-/// work while consumer waits for NFS to drain.
-pub const WRITE_PIPELINE_DEPTH: usize = 16;
+/// Write pipeline depth. 0.21.8 restored the writer-thread + 128 MiB
+/// byte-bounded ring inside WritebackFile, which absorbs kernel
+/// writeback stalls downstream of this channel. The pre-0.21.8
+/// rationale ("smaller buffer reduces backpressure risk when
+/// sync_file_range blocks") no longer applies — `sync_file_range` only
+/// hits the writer thread, never this channel. Doubling 16 → 32 gives
+/// the matroska builder headroom to keep emitting frames when the sink
+/// momentarily lags behind, smoothing out the 5-50 MB/s instantaneous
+/// burst pattern observed on NFS bidirectional workloads at 0.21.8.
+pub const WRITE_PIPELINE_DEPTH: usize = 32;
 
 /// Channel depth for write-through pipelines. Each `send` fully
 /// drains before the next can enqueue. Use this when the producer