feat: add per-model-request token usage tracking for LLM cost visibility

[Kingsuperyzy]

What problem does this PR solve?

This PR introduces per-model-request token usage tracking for model requests in RAGFlow. It intercepts model calls at the middleware layer and persists token consumption data to the database, enabling precise visibility into LLM usage costs at the request level.

Type of change

• New Feature (non-breaking change which adds functionality)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.52%. Comparing base (384fa6f) to head (f82dd96).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #13726      +/-   ##
==========================================
- Coverage   98.11%   96.52%   -1.60%     
==========================================
  Files          10       10              
  Lines         690      690              
  Branches      108      108              
==========================================
- Hits          677      666      -11     
- Misses          4        8       +4     
- Partials        9       16       +7     

☔ 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.

[Lynn-Inf]

I think it would be better to pass the biz_type and biz_id parameters when instantiating LLMBundle. Also, please make sure to use English for comments.

[Kingsuperyzy]

@Lynn-Inf
I’ve made the following updates based on your suggestions.

1. Pass biz_type, biz_id, and session_id parameters when instantiating LLMBundle across all production code paths.
2. Ensure all comments in modified files are written in English.