Add --n-cpus to parallize cubids validate --sequential

[tien-tong]

Closes #475 and #477.

Changes proposed in this pull request

1. For cubids validate:

change --sequential to --validation-scope {dataset, subject} (default: dataset)
change --sequential-subjects to --participant-label

2. Parallel validation for --validation-scope subject (cubids/workflows.py)

• Added parallel processing with ProcessPoolExecutor
• Implemented _validate_single_subject() to process one subject per process
• Used hardlinks -> symlinks -> copy fallback to reduce I/O
• Introduced n_cpus and max_workers; default worker count derived from n_cpus

3. Tests (cubids/tests/)

• Test in test_cli.py for validation with --validation-scope subject --n-cpus
• Test fixes in test_cubids.py (replace comments with assert)

Documentation that should be reviewed

docs/example.rst

• Noted --validation-scope subject l avoids "RangeError: Invalid string length" on large datasets
• Added example: cubids validate BIDS_Dataset_DataLad v0 --validation-scope subject --n-cpus 4

[tien-tong]

For cubids validate:

• change --sequential to --validation-scope {dataset, subject} (default: dataset)
• change --sequential-subjects to --participant-label

[mattcieslak]

Looks good! In the future it might be nice to break up the function into smaller functions, but as long as tests are passing this is good for now

[mattcieslak]

we will want to write more complex tests here eventually. This is fine for now