feat(maya): add Maya Research TTS plugin

[saicherry93479]

Summary

• Adds livekit-plugins-maya (livekit-plugins/livekit-plugins-maya/), a WebSocket TTS provider covering ten Indian languages plus Indian English, with every voice speaking all eleven. Connections come from utils.ConnectionPool, so one socket carries a whole conversation and the handshake is paid once rather than per utterance.
• Maya's v2 protocol needs a handshake per connection, so _connect_ws sends {"type": "start", "v2": true, ...} and waits for the metadata reply before returning the socket — turns sent before that are rejected rather than served in the older frame shape, so the pool must never hand out a socket that hasn't completed it. A rejected handshake raises APIError instead of yielding one that would fail every turn.
• Each agent turn maps to one Maya context: SynthesizeStream mints a context_id, sends each tokenized sentence under it with continue: true, and closes with an empty continue: false frame, yielding exactly one end. Sentences are not gated on the previous one's audio, so a multi-sentence reply streams continuously.
• A turn that ends without its terminator — the caller interrupted, or the task was cancelled — is dropped with a targeted cancel before the socket returns to the pool. Without it the server keeps generating and the next borrower of that connection receives audio from a dead turn.
• Voice, language and model are passed through as given rather than checked against a fixed list; Maya validates them and answers with an error frame, so values it adds later work without a release here.
• ChunkedStream runs over the same websocket rather than Maya's HTTP endpoint, so the plugin has one transport and one auth path.
• Registers the workspace member, the maya optional dependency, and an entry in the cross-provider SYNTHESIZE_TTS suite. version.py tracks the livekit-agents version (1.6.10), matching the sibling plugins so the livekit-plugins-maya>=1.6.10 pin resolves.

Example usage

from livekit.plugins import maya

tts = maya.TTS(voice="Ananya", language="hi")

The API key is read from MAYA_API_KEY.

Test plan

• uv run pytest tests/test_plugin_maya_tts.py — 25 tests against a local server speaking the v2 protocol, covering the handshake contents, the turn model, connection reuse across turns, cancel-on-interrupt, no cancel for a turn that finished, a rejected handshake, and error frames
• uv run ruff check && uv run ruff format --check
• uv run mypy livekit-plugins/livekit-plugins-maya/livekit/plugins/maya/
• Live against the production endpoint: one-shot synthesis, multi-sentence turns sharing one context, consecutive turns reusing the socket, and repeated barge-in followed by a clean turn with no leftover audio

Happy to adjust anything to match house style.

[CLAassistant]

All committers have signed the CLA.