The upstream _parse_chat_history enforcement code uses a first_fc_seen
flag that only adds DUMMY_THOUGHT_SIGNATURE to the first function_call
without thought_signature. Parallel function calls (position 2+) remain
unpatched, causing Gemini API 400 errors for all Gemini 2.5+ models.
Additionally, _is_gemini_3_or_later only matches 'gemini-3', missing
Gemini 2.5 models entirely.
This patch:
1. Extends _is_gemini_3_or_later to also match gemini-2.5 models
2. Wraps _parse_chat_history to ensure ALL function_call parts in ALL
model messages have thought_signature (not just the first one)
- Add episode_group (g=) parameter parsing to explicit media tags in both Python and Rust metainfo parsers
- Propagate episode_group through MetaInfo, MetaBase, MediaInfo, and context models
- Update SKILL.md and update_custom_identifiers.py docs to describe episode group usage
- Add tests for episode_group recognition in metainfo and chain recognition logic
- Introduce RUST_ACCEL config to control all rust fast paths
- Fallback to Python implementations when disabled, preserving filter semantics
- Expose rust acceleration status in system info API
- Update CLI docs to reflect new toggle
- Add tests for runtime switch and fallback behavior
- Simplify and centralize logic for filtering TV episodes during best_version (wash) mode, ensuring only episodes with strictly higher priority are considered for upgrade.
- Always update subscribe.note with downloaded episodes regardless of best_version state, ensuring download history is reliably tracked and available for all subscription modes.
- Remove redundant episode_group field from subscribe dict output.
- Refactor search: remove multi-page search logic, streamline concurrent site search for both sync and async paths, and update progress reporting accordingly.
- Remove obsolete tests for allowed_episodes propagation and note tracking, as logic is now unified and simplified.
