fix(media-use): forward Kokoro speech speed

[miguel-heygen]

Summary

• forward the audio request speed value to the local Kokoro hyperframes tts subprocess
• centralize Kokoro CLI argument construction
• add regression coverage for a non-default 1.15 speed

Reproduction

A media-use audio request with provider: kokoro and speed: 1.15 reaches synthesizeOne, but the spawned command omits --speed; generated lines therefore retain the same duration as speed 1.0. The underlying hyperframes tts command already supports --speed.

Verification

• node --test skills/media-use/audio/scripts/lib/tts.test.mjs skills/media-use/audio/scripts/lib/tts.spawn.test.mjs (21 tests)
• npm run test:skills (230 tests)
• oxfmt and oxlint on changed files
• git diff --check

[miguel-heygen]

Consolidated with the related TTS provider and error-handling fixes in #2407. The still-relevant behavior from the retired skill path is ported into media-use with regression coverage.