Use valid repeated-measures inference for conditional IC

[stefan-jansen]

Closes #59

Summary

• stop returning a fabricated significance p-value for ungrouped inputs that provide only one pooled IC per quantile
• compute matched per-date IC observations for panel inputs and compare conditioning quantiles with a Friedman repeated-measures test
• preserve descriptive IC range and variation when inference is unavailable

Reproduction and verification

• before the fix, independent arrays produced 40 p-values below 0.05 across 100 deterministic seeds
• uv run pytest tests/test_evaluation/test_conditional_ic.py -q -n 0 (29 passed)
• uv run ty check
• pre-commit run --all-files
• uv run pytest tests/ -q -n auto --timeout 120 (5342 passed, 75 skipped)

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.