fix(meetings): validate reaction relay route binding

[mickelr]

Description

• Validate incoming reaction relay events against the active LLM binding using event.headers.route.
• Prefer the practice-session LLM binding when the practice-session LLM connection is active, matching the Voicea publish transport pattern.
• Add unit coverage for default-session and practice-session route match/mismatch cases.

Testing

source ~/.nvm/nvm.sh && nvm use 22.14 && yarn workspace @webex/plugin-meetings test:unit --targets meeting/index.js

Focused #processRelayEvent block passed: 7 passing.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d04eb826e1

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Validate against the event's session binding

When the practice-session socket is connected, this always compares every relay event to the practice binding, even events delivered on the default LLM channel. Mercury tags each incoming envelope with its sessionId and emits default events on event:relay.event while practice events use event:relay.event:llm-practice-session; because both listeners remain registered, a default-session reaction with headers.route === llm.getBinding() will now be skipped whenever the practice channel is also connected. Use the event/session that delivered the relay event to choose the binding instead of global practice connectivity.

Useful? React with 👍 / 👎.