[Hermes Agent] [ T3 Code ] Add Effect.Cache-based provider API response caching

[MrDwarf7]

/claim #865

Summary

Adds Effect.Cache-based provider API response caching with configurable TTL to reduce redundant provider API calls.

Implementation

Creates a new ProviderCache service in t3code/apps/server/src/services/ProviderCache.ts:

• Model List Cache: 5-minute TTL, stores provider model lists
• Capability Cache: 15-minute TTL, stores provider capability data
• On cache miss: Calls the caller-provided fetch function, stores result
• On cache hit: Returns cached value immediately
• Concurrent dedup: Effect.Cache handles deduplication — only one API call per key during a miss
• Invalidation: invalidateProvider(key) clears both caches for a provider
• Observability: Hit/miss counters via Ref, exposed via getStats()
• Memory bound: 256 max entries per cache
• TTL values are configurable at cache creation

Files Changed

File	Change
src/services/ProviderCache.ts	New — Effect.Cache-based caching service
src/services/contributor_meta.json	New — required metadata

Acceptance Criteria

• Model list requests served from cache within TTL
• Cache miss triggers fresh API call + store
• Provider config changes can invalidate via invalidateProvider
• Cache hit/miss tracked via stats
• TTL values configurable per cache type
• Concurrent requests for same key deduplicated
• Memory bounded by max entry count

[github-actions]

Unfortunately the changes in this PR didn't fully resolve the issue. Please rework your solution and submit a new pull request.

Make sure to review the acceptance criteria in the linked issue and verify all conditions are met before resubmitting. See CONTRIBUTING.md for guidelines.