feat(tts): add speed support for Edge TTS, OpenAI TTS, and MiniMax global fallback

[teknium1]

Summary

Salvage of PR #7378 by @0xbyt4 — cherry-picked onto current main with fixes.

Adds a speed config parameter for TTS providers. Users set tts.speed: 1.5 in config.yaml for a global default, or override per-provider with tts.edge.speed, tts.openai.speed, etc.

Changes from original PR

1. Fixed config precedence — provider-specific speed now correctly overrides global (original had it backwards)
2. Wired MiniMax into global fallback — MiniMax previously only read tts.minimax.speed; now falls back to tts.speed like Edge and OpenAI
3. Added 12 regression tests covering precedence, defaults, Edge SSML rate conversion, and OpenAI clamping

Config example

tts:
  provider: edge
  speed: 1.5           # global default
  edge:
    speed: 2.0         # overrides global for Edge only

Provider details

• Edge TTS: converts speed float to SSML prosody rate (+50%, -25%, etc.)
• OpenAI TTS: passes speed param clamped to 0.25–4.0
• MiniMax: passes speed in voice_setting payload

Test plan

tests/tools/test_tts_speed.py — 12 passed
tests/tools/test_tts_mistral.py — 17 passed (no regression)

Closes #7378 — credit to @0xbyt4 for the original implementation.

[github-actions]

⚠️ Supply Chain Risk Detected

This PR contains patterns commonly associated with supply chain attacks. This does not mean the PR is malicious — but these patterns require careful human review before merging.

⚠️ WARNING: base64 encoding/decoding detected

Base64 has legitimate uses (images, JWT, etc.) but is also commonly used to obfuscate malicious payloads. Verify the usage is appropriate.

Matches (first 20):

205:+            aes_key_b64=base64.b64encode(aes_key).decode("ascii"),

---

Automated scan triggered by supply-chain-audit. If this is a false positive, a maintainer can approve after manual review.