docs: library comments say what the scanner and resolver do, not how they came to be (#6277)

Rewrites the internal/library comment blocks to rule 32. Task, milestone,
decision and rule ids are gone; the incidents behind them are restated as
the constraint they left (why a file is deleted before its row, why the
missing mark is a mark, why duplicates block on duration). Counts and
album names from one library are dropped from production comments; test
comments keep the names their fixtures use. The tagReadVersion log keeps
each version's before and after, without the measurements.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

internal/library/loudness.go

@@ -20,7 +20,7 @@ import (
 	syncpkg "git.fabledsword.com/bvandeusen/minstrel/internal/sync"
 )
 
-// Loudness analysis (M464 #4995).
+// Loudness analysis.
 //
 // Every track is measured with ffmpeg's EBU R128 filter, which reports the
 // ITU-R BS.1770 values loudness normalization works from:
@@ -39,7 +39,7 @@ import (
 // 100 ms). Those are kept as a histogram, because an album's loudness is the
 // gated loudness of every block on the album, not an average of its tracks'
 // values. With the histograms stored, album loudness is exact and needs no
-// second decode (#4996).
+// second decode.
 
 // loudnessVersion stamps how a track_loudness row was measured. Bump it when
 // the measurement changes (the filter's options, the histogram's bins) and the
@@ -136,7 +136,7 @@ func (h blockHistogram) empty() bool { return len(h.counts) == 0 }
 // loudness of the blocks that passed it. ok is false when nothing passes.
 //
 // The same function measures an album, over the sum of its tracks'
-// histograms (#4996).
+// histograms.
 func (h blockHistogram) gatedLoudness() (lufs float64, ok bool) {
 	energy := func(i int) float64 {
 		l := loudnessHistFloor + float64(int(h.start)+i)/loudnessHistPerLU
@@ -354,7 +354,7 @@ const (
 // value from an older method is a better gain than none until it is redone.
 //
 // A stored measurement is logged to the sync feed as a track upsert, so
-// clients that cache the library (Android) pick up the new gain (#4997). If
+// clients that cache the library (Android) pick up the new gain. If
 // the log fails nothing is stored, and the backfill tries the track again.
 func storeLoudness(
 	ctx context.Context, db dbq.DBTX, logger *slog.Logger,
@@ -394,7 +394,7 @@ func storeLoudness(
 			}
 		}
 	}
-	// Logged before the write, as reconcile does (#2704): a log that then
+	// Logged before the write, as reconcile does: a log that then
 	// finds the write failed costs clients one wasted re-read, but a write
 	// that lands with no log is never retried, since the track is settled, and
 	// clients would never learn its gain.