fix: fix L01 encoding and decoding

[Lash-L]

Fix L01 on the new devices api

[Copilot]

Pull Request Overview

This PR fixes L01 protocol encoding and decoding in the new devices API by properly handling protocol version information in local communication channels.

Key Changes:

• Updated encode_message to use LocalProtocolVersion enum instead of plain strings for better type safety
• Added protocol_version property to LocalChannel to expose the negotiated protocol version
• Modified local RPC channel creation to pass the correct protocol version when encoding messages

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
roborock/protocols/v1_protocol.py	Changed version parameter type from str to LocalProtocolVersion enum and updated encoding to use .value.encode()
roborock/devices/v1_rpc_channel.py	Updated create_local_rpc_channel to pass the negotiated protocol version from local_channel.protocol_version
roborock/devices/local_channel.py	Added protocol_version property that returns the negotiated version or defaults to V1; also fixed callback assignment in _update_encoder_decoder

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Lash-L]

I can move the responsibility for this further upstream if you'd like.

[allenporter]

i generally get the idea of caching things, but i'm having a hard time seeing how often this will get used. Are we creating local channels repeatedly? i would have thought once per session given once connected it will stay connected.

[allenporter]

(I would cache this like or in NetworkInfo, but i realize it could change if firmware version changes so i casee thats difficult)

[Lash-L]

Right now, seemingly we are. On every reconnect. Rob was noticing that it took longer for his system to reconnect on disconnects

[allenporter]

Are we reconnecting so often that this matters in practice? Ideally we reconnect at most once per day so presumably reconnect is not the primary cost. I can see wanting to optimize startup time, though, which i don't think this does. I guess I think if we really want to cache this to improve performance then we should also cache the same way we cache network info to avoid the extra round trip on startup too?

[allenporter]

I think we should come back to the protocol cache in a separate PR, so its reverted for now.

[allenporter]

Offline Lash-L also reviewed and gave an LG -- just for audit log :) (not sure how approval works on two party CL)