feat: add dataset, prepare_data pipeline and fix McGill converter

- src/dataset.py: ChordDataset wrapping .pt files with pad/truncate
- scripts/prepare_data.py: tokenize .chord to .pt with train/val/holdout
  split, logs token length stats and style/function distributions
- src/external_converters/mcgill_to_chord.py: rewrite parser for real
  McGill v2 format (2-column annotation, each bar in its own pipe group,
  interval bass notation e.g. /5 and /b3)
- .gitignore: exclude data/processed/train, val, holdout subdirectories
- tests: 37 new tests for ChordDataset and converter (260 total, all pass)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/fixtures/mcgill_test/0001/salami_chords.txt

@@ -3,13 +3,7 @@
 # metre: 4/4
 # tonic: C
 
-0.000000	Z
-4.000000	A,verse	C:maj
-8.000000	.	F:maj
-12.000000	.	G:7
-16.000000	.	C:maj
-20.000000	B,chorus	F:maj
-24.000000	.	C:maj
-28.000000	.	G:7
-32.000000	.	C:maj
-36.000000	Z
+0.000000	silence
+4.000000	A, verse, | C:maj | F:maj | G:7 | C:maj |
+20.000000	B, chorus, | F:maj | C:maj | G:7 | C:maj |
+36.000000	silence