fix: prevent model mixing between LiteLLM configuration profiles

[roomote]

This PR attempts to address Issue #9169. Feedback and guidance are welcome.

Problem

When two configuration profiles are created with LiteLLM API provider but different instances (different base URLs), the list of models was always pulled from one of them, causing issues when they don't provide the same models.

Root Cause

The model cache was using only the provider name as the cache key, not distinguishing between different LiteLLM instances with different base URLs.

Solution

• Added unique cache keys that include a hash of the base URL for providers that can have multiple instances (LiteLLM, Ollama, LM Studio, Requesty, DeepInfra)
• Updated file cache naming to include instance-specific information
• Improved cache flushing to handle all cache keys for a provider

Testing

• All existing tests pass
• The fix ensures that each LiteLLM instance maintains its own separate model cache

Fixes #9169

---

Important

Fixes model cache issue by using unique cache keys with base URL hash for LiteLLM and similar providers in modelCache.ts.

• Behavior:
  • Fixes model cache issue for LiteLLM and similar providers by using unique cache keys with base URL hash in modelCache.ts.
  • Ensures each instance maintains its own model cache, preventing model mixing.
• Functions:
  • Adds getCacheKey() and getCacheFilename() to generate unique cache keys and filenames.
  • Updates writeModels(), readModels(), and getModels() to use new cache key logic.
  • Modifies flushModels() to clear all cache keys for a provider.
• Testing:
  • All existing tests pass, confirming separate model caches for each instance.

^{This description was created by for 2081cb1. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Rooviewer   See task on Roo Cloud

Review completed. Found 2 issues that should be addressed:

• Add "roo" provider to the cache key differentiation list (line 43) to prevent model mixing for multiple Roo instances
• Update flushModels to also clear file cache entries, not just memory cache

_{Mention @roomote in a comment to request specific changes to this pull request or fix all unresolved issues.}

[roomote]

The "roo" provider is missing from this list but also supports optional baseUrl (see line 166 in src/shared/api.ts). When users configure multiple Roo provider instances with different base URLs, their model caches will incorrectly share the same key, causing the same model mixing issue this PR aims to fix.

_{Fix it with Roo Code or mention @roomote and request a fix.}

[roomote]

The flushModels function only clears the in-memory cache but doesn't remove corresponding file cache entries (created by writeModels at line 64). When a user flushes models for a provider with multiple instances, stale file caches will persist and could be read back by readModels (line 72), defeating the purpose of the flush operation. Consider iterating through cache files matching the provider pattern and deleting them, similar to how the memory cache keys are being cleared.

_{Fix it with Roo Code or mention @roomote and request a fix.}