feat: add redis response caching

[johnpanos]

Description

Adds Redis-based response caching to reduce latency and costs for repeated LLM requests. Our NLP team needed configurable caching with standard HTTP semantics for cache control.

Features:

• Shared Redis cache across all ext_proc instances
• HTTP Cache-Control support (no-cache, no-store, private, max-age)
• Per-route TTL configuration (default: 1 hour)
• x-aigw-cache: hit/miss response header for observability

Configuration:

• responseCache field on AIGatewayRoute
• extProc.redis.addr in Helm values (or secretRef for production)
• Controller flags for Redis connection

Related Issues/PRs

N/A

Notes for reviewers

Main changes:

• internal/cache/ - Cache interface, Redis client, Cache-Control parsing
• internal/extproc/processor_impl.go - Cache lookup/store logic
• api/v1alpha1/ai_gateway_route.go - ResponseCacheConfig API
• examples/response-cache/ - Example manifests
• site/docs/capabilities/traffic/response-caching.md - Documentation
• tests/e2e/response_cache_test.go - E2E tests

[codecov-commenter]

Codecov Report

❌ Patch coverage is 95.58011% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.20%. Comparing base (fefb039) to head (d6357f5).

Files with missing lines	Patch %	Lines
internal/extproc/processor_impl.go	93.26%	3 Missing and 4 partials ⚠️
internal/cache/cache.go	96.96%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1803      +/-   ##
==========================================
+ Coverage   84.04%   84.20%   +0.15%     
==========================================
  Files         117      120       +3     
  Lines       12990    13171     +181     
==========================================
+ Hits        10917    11090     +173     
- Misses       1418     1422       +4     
- Partials      655      659       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[johnpanos]

We've been running this in prod for around a few days now, and so far it's been holding up really well.

[missBerg]

Might be too much of a stretch for initial implementation, but I'm thinking if this could be done in a way that it would work for pure Envoy Gateway 🤔 so that AIGW simply utilizes it with a different cache key pattern.

So that people needing caching for non-inference could also use it.

Appreciate this may be too much for initial implementation but would be sooo cool!