feat(lattice): add miscalibration detector using Brier score

[Aqsa30nz]

Summary

Implements the session-level MiscalibrationDetector for the v2 Cognitive Lattice.

Fixes #661

Changes

• Added CalibrationEntry to store confidence/outcome pairs.
• Added MiscalibrationDetector for tracking session-level calibration.
• Implemented Brier Score calculation.
• Added configurable warning and blocking thresholds.
• Added evaluate() returning a MiscalibrationResult.
• Exported the new lattice components via agentwatch.lattice.
• Added unit tests covering:
  • empty history
  • recording predictions
  • invalid confidence values
  • well-calibrated sessions
  • warning threshold
  • blocking threshold
  • sample size reporting

Testing

pytest tests/test_miscalibration.py -v
pytest tests/test_shadow_filesystem.py tests/test_miscalibration.py -v

Both test suites pass successfully.

[SHAURYASANYAL3]

This miscalibration detector looks like a solid addition to the V2 Cognitive Lattice. However, there are merge conflicts. Please resolve them and update the PR so I can review the final diff.