AI modules (ai_summarizer.py, ai_providers.py) lack unit tests

[shenxianpeng]

Description

Core AI modules have zero test coverage:

• gitstats/ai_summarizer.py (~350 lines) — contains complex logic for data preparation, caching, summary generation
• gitstats/ai_providers.py (~280 lines) — contains 4 providers (OpenAI, Claude, Gemini, Ollama), factory class, and retry logic

Proposed Fix

Add the following test cases:

ai_summarizer.py

• prepare_index_data(): verify correct extraction of stats data, bot filtering
• prepare_activity_data(): verify peak activity detection
• prepare_authors_data(): verify bot filtering and author sorting
• prepare_lines_data(): verify line stats calculation
• _filter_human_authors(): verify recognition of various bot patterns
• _get_cache_key(): verify cache key generation
• generate_summary(): mock AI provider, verify cache hit/miss logic

ai_providers.py

• AIProviderFactory.create(): known/unknown provider names
• _get_model_with_fallback(): edge cases
• _retry_with_backoff retry logic for each provider

Affected Code

• gitstats/ai_summarizer.py
• gitstats/ai_providers.py