[WIP] feat(llm): migrate to official SDKs and add use_max_completion_tokens…

[stay-foolish-forever]

Description

Migrate LLM client from custom HTTP implementation to official SDKs:

• github.com/openai/openai-go v3.38.0
• github.com/anthropics/anthropic-sdk-go v1.46.0

Key changes in this branch:

1. SDK Migration (refactor)

   • Use SDK native streaming with RawJSON() for format preservation
   • ExtraBody support via option.WithJSONSet
   • Custom User-Agent headers via option.WithHeader
   • Retry logic delegated to SDK (option.WithMaxRetries)
   • StreamCompletion now accepts context.Context for cancellation
   • URL normalization reversed (strip suffixes, SDKs append them)
   • Anthropic tool schema includes required field constraints
   • OpenAI streaming: include_usage enabled for token stats
   • Remove unused resolveUsage/probePath from usage_resolver.go

2. Token Usage Fix

   • Compute totalTokensUsed as input+output instead of relying on API TotalTokens, which some providers report inconsistently

3. use_max_completion_tokens Config Option (feat)

   • Add UseMaxCompletionTokens field to ResolvedEndpoint and ClientConfig
   • Support config file field: llm.use_max_completion_tokens
   • Support environment variable: OCR_USE_MAX_COMPLETION_TOKENS
   • Wire into CLI config set command for user toggle
   • Default to max_tokens for backward compatibility

4. Test Coverage

   • Comprehensive unit tests for client.go (message conversion, utility functions, constructors, streaming response handling)
   • Unit tests for setConfigValue covering all config keys (30 cases)

5. Documentation

   • Add llm.use_max_completion_tokens config key and OCR_USE_MAX_COMPLETION_TOKENS env var to both EN and ZH READMEs

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Refactoring (no functional changes)
• Documentation update
• CI / Build / Tooling

How Has This Been Tested?

• make test passes locally
• Manual testing (describe below)
  Tested using various LLM model provider to ensure it functions correctly

Checklist

• My code follows the project's coding style (go fmt, go vet)
• I have performed a self-review of my code
• I have added tests that prove my fix is effective or my feature works
• New and existing unit tests pass locally with my changes
• I have updated the documentation accordingly (if applicable)
• I have signed the CLA

Related Issues

Issue #30