fix(whatsapp): send voice replies as native audio attachments

[sqsge]

Summary

This PR adds a WhatsApp-specific send_voice() override so voice replies are sent through the existing native media path instead of falling back to the base text placeholder behavior.

Why

The WhatsApp bridge already supports sending audio media payloads.

However, the Python WhatsApp adapter did not override send_voice(), so voice replies fell back to the base platform implementation, which sends a text placeholder like 🔊 Audio: <path> instead of a native media attachment.

This change stays intentionally narrow and wires the missing adapter method into the existing WhatsApp media-send path.

Changes

• add a thin send_voice() override to WhatsAppAdapter
• route voice replies through _send_media_to_bridge(chat_id, audio_path, "audio", caption)
• add regression coverage to verify that send_voice() forwards the correct media type and caption

Fixes #9236

Testing

I added a regression test in tests/gateway/test_whatsapp_connect.py covering the adapter behavior directly.

In this local environment, the repository test suite for this file could not be run end-to-end because pytest-asyncio and some default pytest addopts dependencies are not installed here.

To reduce that risk, I verified the change with:

• a direct Python-level smoke check confirming send_voice() forwards to the "audio" media path
• a focused regression test covering the forwarding behavior and caption passthrough

[teknium1]

Thanks for the fix! The same send_voice() override already landed on main in commit ed76185 ("feat(whatsapp): implement send_voice for audio message delivery") via #12656, so this PR is now a no-op diff against current main. Closing as superseded — the behavior you fixed is in main today.

ed76185

[alt-glitch]

Likely duplicate of #13002 — same WhatsApp send_voice() fix already merged.

[alt-glitch]

Likely duplicate of #13002