BOT: Fix #811: batch failed metric warnings into a single message

[nikosbosse]

Summary

• When multiple metrics fail during score(), apply_metrics() now collects all errors and emits a single batched cli_warn() summarizing all failures, instead of warning separately for each failed metric
• run_safely() no longer calls cli_warn() directly — it returns error info as a structured list, and apply_metrics() handles the aggregated warning
• Fixes Should score()/apply_metrics() return warning about failed metrics in a batch? #811

Root cause

run_safely() called cli_warn() immediately inside the lapply loop for each metric that failed. With N failing metrics, users saw N separate warnings with no clear summary.

What the fix does

1. run_safely() now returns list(result = NULL, error = <message>) on failure instead of warning
2. apply_metrics() collects all failures during the loop, then emits a single cli_warn() at the end listing all failed metrics and their error messages
3. Existing tests in test-class-forecast-quantile.R updated to match the new warning format

Test coverage added

• apply_metrics() emits a single batched warning when multiple metrics fail
• apply_metrics() emits no warning when all metrics succeed
• Batched warning message includes metric name and error details for each failure
• score() emits a single batched warning when multiple metrics fail
• run_safely() returns error info instead of warning directly
• Existing one-sample score() test updated for new format
• Full test suite passes (706 tests), R CMD check: 0 errors, 0 warnings

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.84%. Comparing base (ac0c01a) to head (d99c867).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1085      +/-   ##
==========================================
+ Coverage   97.83%   97.84%   +0.01%     
==========================================
  Files          35       35              
  Lines        1845     1854       +9     
==========================================
+ Hits         1805     1814       +9     
  Misses         40       40              

☔ 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.

[nikosbosse]

CLAUDE: Approve — Clean, well-scoped fix that correctly moves warning emission from per-metric (run_safely) to batched (apply_metrics). The implementation is minimal: run_safely() returns a structured error list instead of calling cli_warn() directly, and apply_metrics() collects all failures then emits a single cli_warn() summarizing all failed metrics. All 6 test specifications are covered with thorough assertions. Existing tests in test-class-forecast-quantile.R are correctly updated to match the new batched format. No issues found.