fix: add Redis graceful degradation and health check (#126)

[Xuccessor]

Summary

AdaptiveRateLimitGuard previously threw an unhandled exception when Redis was unavailable, taking down all endpoints. The health check also had no visibility into Redis status. This PR adds fail-open degradation to the rate limit guard and a Redis ping check to the readiness probe.

Changes

• adaptive-rate-limit.guard.ts: Wrap getOrThrow() and Redis operations in try/catch — logs a warning and allows requests through when Redis is down
• health.service.ts: Add checkRedis() method using RedisService.getOrThrow().ping(); include redis key in ReadinessResponse.checks; update ReadinessResponse interface
• health.module.ts: Import RedisModule to make RedisService injectable in HealthService
• health.controller.spec.ts: Add RedisService mock and update assertions to include redis in readiness checks

Testing

Build and lint pass. All 4 health controller tests pass. When Redis is unavailable, rate limiting is skipped rather than throwing 500 errors.

Closes #126

[kilodesodiq-arch]

LGTM