fix(tooling): coverage.yml prints per-layer FILE COUNTS beside real percentages, and pipes prose into GITHUB_OUTPUT

[koniz-dev]

Context

Two small pieces of confusion in coverage.yml. Neither breaks the gate - the
per-layer thresholds are computed and enforced correctly (see the closing comment
on #25) - but the first one directly cost a bogus P2 issue, so it is worth
removing.

1. The metrics step prints file counts that read like coverage

📊 Coverage Metrics:
  Overall: 89.8%
  Domain files: 15
  Data files: 12
  Presentation files: 11
  Core files: 70

15 is a count of SF: records, not a percentage, sitting directly under a real
percentage under a heading that says "Coverage Metrics". The real per-layer
figures are produced by a different step (layer_coverage) and printed
elsewhere. I read this block, concluded the thresholds were not enforced, and
filed #25 on that basis. It is a trap for the next reader too.

2. calculate_layer_coverage.sh emits prose into $GITHUB_OUTPUT

The script prints label lines between its key=value pairs:

Output values with % suffix for display
domain_display=100.0%
...
Output numeric values for comparison
domain=100.0
...
Summary of coverage by layer
Domain Layer: 100.0%

The workflow appends every line to $GITHUB_OUTPUT with
while IFS='=' read -r key value, so a line with no = becomes
Output values with % suffix for display=. Harmless today, but it is writing
junk keys into a step output and would silently break if a label ever contained
=.

Acceptance criteria

1. Calculate coverage metrics no longer prints per-layer file counts under a
   "Coverage Metrics" heading. Either it reports the real per-layer percentages
   (reusing steps.layer_coverage.outputs.*) or it drops the per-layer lines and
   reports overall only.
2. Anything still emitting a file count says so unambiguously in its label
   (e.g. "files measured", not "Domain files" under "Coverage Metrics").
3. scripts/test/calculate_layer_coverage.sh separates machine-readable output
   from human output: only key=value lines reach $GITHUB_OUTPUT. A
   --github-output flag, or sending prose to stderr, both work.
4. $GITHUB_OUTPUT contains no keys with spaces after the change. Show the
   step's parsed output.
5. The Enforce coverage thresholds step still gates on the same five numbers
   and still passes on main: a workflow_dispatch run concludes success with
   the five ✅ lines. Record the run id.
6. ./scripts/dev/audit_template.sh exits 0.

[koniz-dev]

Setting status:blocked: GitHub Actions is not assigning runners for this repo, so the acceptance criteria cannot be verified.

State right now:

• Coverage Analysis dispatch concluded failure at run level with its only job still queued and no steps executed.
• gh run rerun was accepted; the run returned to queued and stayed there.
• An E2E Android (Patrol) run has been queued since 15:25Z with nothing else occupying runners.
• The last run that actually executed was 14:42Z.
• All eight workflows report state: active.

To unblock: nothing in this repository. Once Actions schedules runners again, re-dispatch and the remaining criteria can be checked. No code decision is pending.

[koniz-dev]

Work is complete and locally verified; only CI confirmation is missing.

Done: PR #42. ./scripts/dev/audit_template.sh exits 0. Both script modes verified against a real lcov.info - --github-output emits exactly eight key=value lines and no emitted key contains a space (criteria 1-4, 6).

Outstanding: criterion 5 only - a dispatched coverage.yml run on main showing the five ✅ threshold lines. That needs a runner.