bug(cli): --threshold compares mean score instead of per-test score

[christso]

Problem

--threshold compares the mean score against the threshold, but RESULT: uses per-test pass/fail (score >= 0.8). This produces contradictory output and exit codes:

RESULT: FAIL  (28/31 passed, mean score: 0.927)
Suite score: 0.93 (threshold: 0.80) — PASS     ← exit code 0

The output says FAIL but the exit code is 0. Users expect --threshold 0.8 to mean "each test must score >= 0.8" — matching the per-test requirement.

Root cause

• formatEvaluationSummary() — per-test pass/fail (score >= hardcoded 0.8)
• formatThresholdSummary() — mean score comparison
• Exit code follows threshold (mean-based), not RESULT (per-test)

Fix

PR #885 — --threshold now overrides the per-test score requirement:

• calculateEvaluationSummary() recomputes passed/failed using the threshold
• RESULT line shows the threshold: 28/31 scored >= 0.8
• Exit code matches RESULT verdict
• Removed separate formatThresholdSummary() — one unified output line