feat(schema): Rewrite PHENOTYPE_SUBJECTS_MISSING to run on each phenotype file

[effigies]

This builds on and takes advantage of #2050 to select phenotype files and compare the participant_id column to the dataset-global dataset.subjects.participant_id, which is populated from participants.tsv.

This does two things for validation implementations:

1. It allows the specific phenotype file with extra participants to be identified, since the phenotype participants are not collapsed into one superset.
2. It allows us to avoid calculating the union at validator startup, which could be potentially costly for files with many phenotype files, as well as avoid loading each phenotype file twice (or hold a long-lived cache of all phenotype data).

Note that #2050 is not strictly required in order to use this check, as validating phenotype file names already requires us to have it as a pseudo-datatype. I do think a second use case (with more expected from BEP36) justifies moving this from implementation detail to official spec language.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.71%. Comparing base (254e22e) to head (2c9e882).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2107   +/-   ##
=======================================
  Coverage   82.71%   82.71%           
=======================================
  Files          17       17           
  Lines        1533     1533           
=======================================
  Hits         1268     1268           
  Misses        265      265           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.