Add regression tests for controller emergency brake and anti-windup

[cursor]

Summary

Adds focused regression coverage for high-risk controller safety paths that remained untested on dev, while avoiding duplication of open drafts #17/#20/#22–#34.

Risky behavior now covered

• Stanley emergency brake: near-zero reference speed while ego is still moving overrides weak/zero PID with ego_min_acceleration * c30_emergency_braking_factor
• Stanley emergency gate: ego below c30_emergency_min_moving_velocity does not trigger the override
• Stanley anti-windup + accel floor: stopped ego clears a positive cte_v_sum and refuses further negative acceleration
• Stanley CTE slow-down: large lateral offset while fast dumps acceleration toward max braking
• PurePursuit velocity_pid: same emergency override and stop-floor/anti-windup contract on the shared velocity PID helper

Test files added/updated

• test/c30_control/test_c30_emergency_brake.py (new)

Why these tests materially reduce regression risk

Emergency braking and anti-windup sit on every Stanley/PurePursuit control tick and gate obstacle-stop / end-of-path deceleration. Silently dropping the override leaves a weak PID coasting into a stop request; losing anti-windup/floor pushes the plant through zero velocity. CTE slow-down is the Stanley-specific path that bleeds speed in corners—easy to regress while tuning gains.

Validation

pytest test/c30_control/test_c30_emergency_brake.py test/c30_control/ -v

All 29 control tests passed (6 new). No production code changes.