Add tracking performance metrics

[FlorianPfaff]

Summary

Adds a richer tracking/evaluation metric surface inspired by TrackerComponentLibrary-style performance tooling:

• point-estimate errors: MSE, RMSE, MAE, squared error
• consistency metrics: NEES, ANEES, NIS, ANIS, chi-square bounds and consistency helpers
• finite-set distances: OSPA, MOSPA, GOSPA
• EOT/extent metrics: polygon IoU aliases, Gaussian 2-Wasserstein distance, extent matrix/Wasserstein errors
• track outcome metrics: purity, false-track rate, missed-track rate, first-detection latency, combined outcome summary
• public exports through pyrecest.utils
• regression tests for the new metric families

Tests

Not run locally in this environment. Suggested command:

python -m pytest tests/test_metrics.py tests/test_track_metrics.py