Add MiniMax voice cloning to Media Center

[octo-patch]

Reason: Add MiniMax voice-clone support to the provider-neutral Media Center actions.

Changes

• upload workspace WAV, MP3, and M4A samples to the regional MiniMax file endpoint
• call the MiniMax voice-clone endpoint with validated file, voice, and model fields
• parse provider status codes and reusable voice IDs into the existing Media Center response envelope
• cover global and China endpoints, direct file IDs, multipart uploads, validation, and error responses

Checks

• corepack pnpm --filter ipollowork-server exec bun test src/extensions/media-center.test.ts
• corepack pnpm --filter ipollowork-server typecheck
• corepack pnpm --filter ipollowork-server test
• git diff --check
• node .codex/skills/ipollowork-maintainable-code/scripts/audit-changes.mjs