fix(web): handle TEXT modality for native-audio models (#4206)

[paulxg12]

Summary

Fixes #4206 — native-audio models fail when TEXT modality is explicitly requested via the ADK web server.

Root Cause

The /run_live WebSocket endpoint accepted modalities from query params without validating against model capabilities. When a user requested TEXT modality for a native-audio model (e.g., gemini-live-2.5-flash-native-audio), the model failed since it only supports AUDIO.

Changes

• adk_web_server.py: detect native-audio models in WebSocket handler and override modalities to ["AUDIO"]
• runners.py: extended native-audio check to also correct explicitly-set ["TEXT"] to ["AUDIO"] (previously only handled None case)

Testing

• Native-audio models with no modality specified → AUDIO (unchanged behavior)
• Native-audio models with TEXT modality requested → AUDIO (fixed)
• Non-audio models → respects user's modality choice (unchanged)

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[paulxg12]

I have read the Google CLA and I hereby sign the CLA