iter14: re-engage FileSectorSource 4 MiB readahead buffer

iter13 strace finding: producer thread (mux ISO reader) spends 80%
of wall-clock in state D (NFS RPC wait), doing 207 preads/sec at
103 us each. Pipe is 78 MB/s isolated read; we use ~25 MB/s.

FileSectorSource was using direct per-sector pread (0.21.3 bypass).
That bypass was justified under Phase 2.5 + 0.21.7 producer-poll cap
("32 MiB refill bursts the TCP connection enough to starve the
writer thread"). Both of those conditions are gone now (iter8
baseline: no Phase 2.5, no producer poll cap).

The 4 MiB buffer infrastructure was preserved with #[allow(dead_code)]
in case re-engagement was ever wanted. iter14 just routes the hot
path through buffer_covers + refill + memcpy. Net diff: ~10 LOC of
business logic, 5 unit tests already cover the contract.

Expected: producer's effective read rate jumps from 25 → 60+ MB/s
(matches dd ceiling for 1+ MiB block reads). If consumer side keeps
up, mean mux climbs from 28.7 → 35-50 MB/s. If consumer is now the
cap, we see a clear ceiling around 30-35 and we know where to look
next.

Critically: this change ONLY affects mux-from-ISO. Disc→ISO sweep,
Pass N bad-sector retry, AACS, drive ops, mapfile, recovery — all
untouched (they use DriveSectorSource which is a separate impl).

src/io/file_sector_source/mod.rs

@@ -103,14 +103,11 @@ pub struct FileSectorSource {
     /// Total file size in sectors. Constant after construction;
     /// surfaced via [`SectorSource::capacity_sectors`].
     capacity: u32,
-    /// 0.21.3+: the app-level buffer is no longer touched on the hot
+    /// 4 MiB readahead buffer. iter14 (2026-05-17): re-engaged after
-    /// path (every `read_sectors` is a direct pread). The fields are
+    /// the 0.21.3 bypass. On a file (vs an optical drive) per-sector
-    /// retained so a future per-source-type policy (e.g. a local-disk
+    /// reads pay one NFS RPC per 2048 bytes. The buffer amortises
-    /// source where batched reads ARE beneficial) can re-enable
+    /// those RPCs to one pread per BUF_SECTORS (~2048 sectors).
-    /// buffering cleanly without re-plumbing the struct.
-    #[allow(dead_code)]
     buf: Box<[u8]>,
-    #[allow(dead_code)]
     buf_start_lba: u32,
     buf_len_sectors: u32,
     /// 0.21.6: bytes read since the last DONTNEED drop. Drives the
@@ -167,7 +164,6 @@ impl FileSectorSource {
 
     /// True if `[lba, lba + count)` is wholly inside the current
     /// buffer window. `count == 0` is vacuously true.
-    #[allow(dead_code)]
     fn buffer_covers(&self, lba: u32, count: u32) -> bool {
         if self.buf_len_sectors == 0 {
             return false;
@@ -183,7 +179,6 @@ impl FileSectorSource {
     /// Refill the buffer so it starts at `lba`. Read as many sectors
     /// as we have buffer space AND file capacity for. Caller has
     /// already checked `lba < capacity`.
-    #[allow(dead_code)]
     fn refill(&mut self, lba: u32) -> Result<()> {
         debug_assert!(lba < self.capacity, "refill past capacity");
         // Don't read past EOF — clamp the request to remaining
@@ -226,28 +221,34 @@ impl SectorSource for FileSectorSource {
         if count == 0 {
             return Ok(0);
         }
-        // 0.21.3: bypass the application-level buffer entirely.
+        // iter14 (2026-05-17): route through the 4 MiB readahead
+        // buffer. The 0.21.3 bypass was justified under Phase 2.5
+        // architecture + producer-poll cap (both gone now). With iter8
+        // baseline (no Phase 2.5, no producer poll cap), one pread per
+        // 2048-byte sector forces ~5000+ NFS RPCs/sec; the producer
+        // thread sits in `D` state 80% of the time waiting on RTT.
+        // A 4 MiB window amortises that to one pread per ~2048 sectors
+        // (~11 ms refill at NFS 91 MB/s ceiling).
         //
-        // Empirically the 32 MiB readahead window (0.21.0–0.21.1) and the
+        // Pathological-large request bypasses the buffer (caller asked
-        // 4 MiB shrink (0.21.2) both regressed mux throughput vs the
+        // for more than fits — defensive, sweep doesn't do this).
-        // pre-Phase-1 0.20.7 baseline on NFS bidirectional workloads
+        if count > BUF_SECTORS {
-        // (sweep ~25 MB/s OK; mux dropped from 18 → 7-8 → 5-6 MB/s).
+            let offset = lba as u64 * SECTOR_SIZE as u64;
-        // Direct pread per call lets the kernel's own readahead policy
+            self.file
-        // run, which interleaves naturally with concurrent NFS writes on
+                .seek(SeekFrom::Start(offset))
-        // the same TCP connection.
+                .map_err(|e| Error::IoError { source: e })?;
-        //
+            self.file
-        // Buffer fields are retained (currently unused on this path) so
+                .read_exact(&mut out[..bytes])
-        // any future per-source policy can be reintroduced without
+                .map_err(|e| Error::IoError { source: e })?;
-        // re-plumbing structure. `refill` / `buffer_covers` are kept too
+            self.buf_len_sectors = 0;
-        // (still exercised by the tests so the API contract is locked).
+        } else {
-        let offset = lba as u64 * SECTOR_SIZE as u64;
+            if !self.buffer_covers(lba, count) {
-        self.file
+                self.refill(lba)?;
-            .seek(SeekFrom::Start(offset))
+            }
-            .map_err(|e| Error::IoError { source: e })?;
+            let off_sectors = (lba - self.buf_start_lba) as usize;
-        self.file
+            let off_bytes = off_sectors * SECTOR_SIZE;
-            .read_exact(&mut out[..bytes])
+            out[..bytes].copy_from_slice(&self.buf[off_bytes..off_bytes + bytes]);
-            .map_err(|e| Error::IoError { source: e })?;
+        }
-        self.buf_len_sectors = 0;
 
         // 0.21.6: periodic page-cache eviction on the read side. Without
         // this, an 85 GB streaming ISO read pins the entire file in