#18927: TTS: default to eleven_flash_v2_5, drop eleven_monolingual_v1, add model/voice to tool

TTS: default ElevenLabs model is now eleven_flash_v2_5 (free-tier friendly); deprecated eleven_monolingual_v1 removed from options; TTS tool accepts optional modelId and voiceId. fix #18895 <!-- greptile_comment --> <h3>Greptile Summary</h3> Updated ElevenLabs TTS to use `eleven_flash_v2_5` as the default model (better free-tier compatibility) and removed deprecated `eleven_monolingual_v1`. Added optional `modelId` and `voiceId` parameters to the TTS tool, allowing agents to customize voice output per request. <h3>Confidence Score: 4/5</h3> - Safe to merge with minor documentation inconsistencies - The code changes are straightforward and well-implemented. The default model change aligns with the goal of free-tier friendliness. The new tool parameters follow existing patterns and include proper override handling. However, documentation in multiple files still references the old default model. - No files require special attention _{Last reviewed commit: aac696c} <!-- greptile_other_comments_section --> <!-- /greptile_comment -->