BOT: Fix #193: Add sd_sample() and iqr_sample() sharpness metrics

[nikosbosse]

Summary

• Adds sd_sample() and iqr_sample() as new sharpness metrics for sample-based forecasts
• Both follow the existing mad_sample() pattern: observation-independent, row-wise apply() over the predicted matrix
• Registered as default metrics in get_metrics.forecast_sample() under names "sd" and "iqr"

Fixes #193

Root cause

Only mad_sample() existed as a dedicated sharpness metric. The issue requested standard deviation and IQR as additional spread measures.

What the fix does

• Implements sd_sample(observed = NULL, predicted, ...) using apply(predicted, 1, sd)
• Implements iqr_sample(observed = NULL, predicted, ...) using apply(predicted, 1, IQR)
• Adds both to the default metric list in get_metrics.forecast_sample()
• Full roxygen documentation with examples and @export

Test coverage added

• 11 new test blocks covering:
  • Correctness against known values (constant row, sequential row, mixed row)
  • observed = NULL handling
  • Error on missing predicted
  • Single-row edge case
  • Default metric registration
  • End-to-end score() integration
  • Agreement with reference apply()-based computation

Test plan

• New tests fail before implementation
• New tests pass after implementation (64/64)
• Full test suite passes (706/706)
• R CMD check: 0 errors, 0 warnings, 2 notes (pre-existing)

🤖 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 (0a57905).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1103   +/-   ##
=======================================
  Coverage   97.83%   97.84%           
=======================================
  Files          35       35           
  Lines        1845     1853    +8     
=======================================
+ Hits         1805     1813    +8     
  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: Automated review — APPROVE. Clean, minimal implementation of sd_sample() and iqr_sample() that correctly follows the established mad_sample() pattern. All 11 test specs implemented and passing. No issues found — functions, docs, exports, imports, and metric registration are all correct.