feat(rtc): expose pre-encoded video publishing in Python

[AmitMY]

Python currently accepts only raw video frames, so applications with an existing encoded stream must decode and encode it again to publish a media track. Expose the pre-encoded source already supported by the bundled Rust FFI, matching the Rust/Go publishing path.

Adds EncodedVideoSource, EncodedVideoFrame, encoder feedback, and public encoder/frame-type enums. Existing LocalVideoTrack publishing accepts the new source. No Rust, protobuf, native-library, or runtime dependency changes are required. The README covers codec matching, pacing, disabled simulcast, keyframe/bitrate feedback, and why container bytes and WebM alpha are separate concerns.

Validation:

• Ruff and full SDK mypy pass (excluding the Rust submodule from Python lint/format checks).
• 136 tests pass; 73 tests requiring server credentials are skipped in the offline suite.
• All 12 new tests pass with a locally built LiveKit 1.13.7 server and the bundled FFI 0.12.80, including publishing compressed VP9 and verifying decoded pixels on a subscriber.

First part of a stack; #839 adds incremental WebM demuxing for opaque media tracks and a separate example forwarding intact WebM, including alpha, through the existing byte-stream API.

CI also exposed a pre-existing fork PR checkout failure in the protobuf-generation job. It now checks out the head repository, verifies generated stubs read-only on forks, and retains automatic commits for same-repository PRs.

[CLAassistant]

All committers have signed the CLA.

[devin-ai-integration]

🔍 Devin Review: 1 flag

Not posted on this PR by your GitHub settings — view it in Devin Review. (Configure)