Add sync/async Text-to-Speech (TTS) client

[GISWLH]

Summary

Adds a first-class TTS client for the public documented Voice/TTS REST APIs, addressing #126.

• client.tts.synthesize(...) → POST /v1/tts (sync + async)
• client.tts.list_voices() / client.tts.get_voice(voice_id) → GET /v1/tts/voices[/{id}]
• Typed models (TtsResponse, Voice, AudioTimestamps, OutputFormat, …)
• Client-side validation aligned with public docs (text length, speed, codecs, replace map)
• Handles raw audio bytes and with_timestamps JSON envelopes (supports both [start, end] and {start, end} graph_times shapes)
• Unit tests with mocked HTTP (no live API key): success, 400/401/404/429, list/get voices, validation
• Examples under examples/sync/tts.py and examples/aio/tts.py

API surface

from xai_sdk import Client

client = Client()
voices = client.tts.list_voices()
audio = client.tts.synthesize("Hello!", language="en", voice_id="eve")
audio.write_to_file("hello.mp3")

timed = client.tts.synthesize("Hello!", language="en", with_timestamps=True)
# timed.audio, timed.duration, timed.audio_timestamps

Deferred (intentionally out of scope for this PR)

• WebSocket streaming TTS (wss://api.x.ai/v1/tts)
• Custom voices CRUD (/v1/custom-voices)
• Realtime / Speech-to-Speech

Happy to iterate if maintainers prefer different method names or want streaming in a follow-up.

Test plan

• uv run ruff format --check / uv run ruff check on changed files
• uv run pyright on TTS modules + clients
• uv run pytest tests/tts_test.py tests/sync/tts_test.py tests/aio/tts_test.py -v (35 passed)
• uv run pytest tests/sync/client_test.py tests/aio/client_test.py tests/tts_test.py tests/sync/tts_test.py tests/aio/tts_test.py -q (62 passed)

Fixes #126