ci(test): enforce coverage-based test selection on PRs

[sbryngelson]

What

Flips the coverage-based test selection from shadow to enforced on PR runs:

• test.yml github (GNU/intel) Test step → --only-changes --select-enforce
• common/test.sh self-hosted Phoenix/Frontier → --only-changes --select-enforce

PRs now run only the tests whose recorded coverage overlaps the changed files. The conservative ladder in coverage.py falls back to run-all for anything risky: unavailable file lists (rung 1), macro/codegen/build inputs (rung 2), hand-written .f90 (rung 3), and — critically — a changed .fpp that no test covers (rung 4). The only "run nothing" path is docs/irrelevant-only changes (rung 7).

Safety / backstops (why this is safe to enforce)

• Pre-merge full check: the NVHPC jobs still run --test-all on every PR, so the full suite runs on at least one config before merge — a selector under-inclusion is caught pre-merge, not just post-merge.
• Master backstop: pushes to master run the full suite (SELECT empty), never cancelled per the concurrency config.
• Selection is visible in each PR's test log (Coverage selection: ran N/M …), so pruning can be eyeballed.

Note

This PR itself touches test.yml/test.sh (non-.fpp) → rung 2 → it runs the full suite, so pruning isn't demonstrated here. A follow-up Fortran-only PR will show the enforced prune. Pruning the NVHPC jobs too (for more savings) is a later step once this proves out.

[sbryngelson]

Added a second commit simplifying the flags: --select-enforce now implies selection, so it no longer needs to be paired with --only-changes (which was redundant — --select-enforce did nothing on its own).

• --select-enforce (alone) → select and prune
• --only-changes (alone) → shadow (print selection, run full suite)

CI enforce paths now pass just --select-enforce --changed-files …. Verified locally: --select-enforce alone prunes (81/578 for a bubbles change, no shadow message); --only-changes alone still shows the shadow message. 273 unit tests pass.

[Copilot]

Pull request overview

This PR switches coverage-based test selection on PR runs from “shadow” (log-only) to “enforced” (actually skipping unselected tests), while keeping conservative fallbacks (run-all) when selection would be unsafe.

Changes:

• Treat --select-enforce as sufficient to trigger coverage selection (no need to also pass --only-changes).
• Update CLI help text around coverage-selection flags.
• Update GitHub-hosted and self-hosted CI runners to pass --select-enforce on PRs.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
toolchain/mfc/test/test.py	Runs coverage selection when either --only-changes (shadow) or --select-enforce (enforced) is set.
toolchain/mfc/cli/commands.py	Updates CLI help strings describing shadow vs enforced selection behavior.
.github/workflows/test.yml	Enforces selection on PRs by passing --select-enforce (and still runs full suite on master pushes).
.github/workflows/common/test.sh	Enforces selection on self-hosted PR runs via --select-enforce.

Comments suppressed due to low confidence (1)

toolchain/mfc/cli/commands.py:486

• The changes-branch help text still says it’s only used for --only-changes, but the selection path now also runs when --select-enforce is set (even without --only-changes). Update the help to reflect both flags so users understand it applies to enforced selection too.

        Argument(name="changed-files", dest="changed_files", type=str, default=None, help="Changed-file list (newline-, space-, or comma-separated; from CI paths-filter). Overrides git detection."),
        Argument(name="changes-branch", dest="changes_branch", type=str, default="master", help="Branch to diff against for --only-changes."),
    ],

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 61.31%. Comparing base (574e53d) to head (3eaba1c).
⚠️ Report is 9 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1468   +/-   ##
=======================================
  Coverage   61.31%   61.31%           
=======================================
  Files          72       72           
  Lines       19771    19771           
  Branches     2852     2852           
=======================================
  Hits        12123    12123           
  Misses       5699     5699           
  Partials     1949     1949           

☔ 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.