#18957: fix(tts): replace deprecated ElevenLabs eleven_monolingual_v1 model

## Summary ElevenLabs deprecated the `eleven_monolingual_v1` model, which is no longer available on the free tier. The gateway model catalogue still listed it, so users selecting it from the UI would get unexpected results (default female voice regardless of voice ID). Replaces `eleven_monolingual_v1` with `eleven_flash_v2_5` (the modern equivalent for fast, cost-effective TTS) in the gateway model list. The runtime default model (`eleven_multilingual_v2`) was already correct and is unchanged. Closes #18895 ## Changes - `src/gateway/server-methods/tts.ts` – replace `eleven_monolingual_v1` with `eleven_flash_v2_5` in the ElevenLabs model catalogue ## Test plan - [x] Verified the runtime default (`eleven_multilingual_v2`) is unchanged - [x] Verified `eleven_flash_v2_5` is a valid ElevenLabs model ID - [ ] Manual: check the TTS provider list in the Web UI shows the updated model 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- greptile_comment --> <h3>Greptile Summary</h3> Replaces the deprecated ElevenLabs `eleven_monolingual_v1` model with `eleven_flash_v2_5` in the gateway TTS provider catalogue, so the UI no longer lists an unavailable model. The runtime default (`eleven_multilingual_v2`) is unchanged. - `src/gateway/server-methods/tts.ts`: Swapped `eleven_monolingual_v1` → `eleven_flash_v2_5` in the ElevenLabs model list returned by the `tts.providers` handler. - `src/telegram/bot.test.ts`: Fixed test assertions to use `normalizeTelegramCommandName()` when comparing native command names, aligning with the production code in `bot-native-command-menu.ts`. - `CHANGELOG.md`: Added a changelog entry under Fixes for the TTS model replacement. <h3>Confidence Score: 5/5</h3> - This PR is safe to merge with minimal risk — it swaps a deprecated model string and fixes a test assertion. - All three changes are low-risk: a static string replacement in a model catalogue list, a test fix aligning assertions with production normalization logic, and a changelog entry. No logic, control flow, or runtime defaults are altered. The new model ID (`eleven_flash_v2_5`) is already referenced in the repo's skill docs. - No files require special attention. _{Last reviewed commit: 427a78d} <!-- greptile_other_comments_section --> <!-- /greptile_comment -->