Refactor scan manifest semantics to canonical scans.tsv with long-form entity columns and participant metadata flow

[Copilot]

SPIMpack used datasets.tsv for scan-level rows, which conflicted with BIDS “dataset” semantics and overloaded metadata responsibilities. This PR makes scans.tsv the canonical scan manifest, removes deprecated scan-manifest aliases in alpha, and introduces participant-level metadata ingestion/output via participants.tsv.

• Canonical scan manifest (alpha simplification)

  • Uses scans.tsv as the canonical scan manifest file.
  • Removes deprecated datasets_tsv / datasets.tsv compatibility handling.
  • Resolves scan table paths via manifest-relative path checks.

• Removal of deprecated scan-level fields/aliases

  • Removes scan-level compatibility traces for deprecated names like dataset_id.
  • Keeps scan processing stable via row iteration without legacy dataset-id assumptions.

• Participant-level metadata support

  • Adds optional participants.tsv ingestion with required participant_id.
  • Uses subject as the canonical participant linkage in scans.tsv when participants are supplied.
  • Validates that scan subjects are present in participants.tsv.
  • Extends manifest model to carry participant records and metadata.

• Long-form scan TSV entity headers for readability

  • Uses long-form column headers in scans.tsv: subject, session, sample, acquisition.
  • Keeps short forms internal for BIDS filename construction (sub, ses, acq tokens in paths).

• BIDS output alignment

  • Generates top-level BIDS participants.tsv from participant metadata when provided.
  • Preserves existing scan packaging behavior and sidecar generation.
  • Keeps scan-specific metadata in scan flow while participant attributes remain participant-scoped.

• UI + docs migration updates

  • Renames UI defaults/copy from datasets to scans (scans.tsv, scans_tsv).
  • Updates UI-generated scan rows and labels to use long-form entity names.
  • Updates README schema/examples to reflect canonical scans.tsv and long-form scan headers.

# manifest.yml (canonical form)
dataset_description:
  Name: Demo
  BIDSVersion: 1.9.0
  DatasetType: raw
  License: CC0
scans_tsv: scans.tsv
participants_tsv: participants.tsv  # optional

[akhanf]

@copilot: the subject column and participant_id column now duplicate each other.

[Copilot]

@copilot: the subject column and participant_id column now duplicate each other.

Addressed in c94d42c (behavior/UI/tests) and 28e982a (README alignment). sub is now the canonical participant linkage in scans, and legacy participant_id is optional compatibility-only.

[akhanf]

@copilot: don't worry about keeping backwards compatibility for deprecated variables/names, this code is still alpha and can have major api changes.. you can just remove any trace of things like dataset_id. also would be good to replace the shorthand names for the column headers (eg sub -> subject, ses -> session, acq -> acquisition), keeping track of the short and long form names so the short ones can still be used when constructing filenames, but long ones used for column headers to improve readability.

[Copilot]

@copilot: don't worry about keeping backwards compatibility for deprecated variables/names, this code is still alpha and can have major api changes.. you can just remove any trace of things like datas...

Done in af4b20e (+ follow-up fixes 84f6ffe, 3b6990d, 44891df). Deprecated scan aliases/compatibility paths were removed, and scans headers now use long names (subject, session, sample, acquisition) while short forms remain internal for BIDS filename construction.