fix(telegram): apply _effective_update_message to _handle_media_message

[EloquentBrush0x]

Summary

PR #28531 introduced _effective_update_message and applied it to
_handle_text_message, _handle_command, and _handle_location_message
so that Telegram channel posts (update.channel_post; update.message == None)
are routed correctly. _handle_media_message was missed.

Bug: the if not update.message: return guard at the top of
_handle_media_message exits immediately for every channel post, silently
dropping all channel media (photos, videos, audio, documents, stickers).

Fix: resolve msg via _effective_update_message at the top of the
handler and remove the now-redundant msg = update.message assignment.
Identical to the pattern already applied to the three sibling handlers.

Changes

• gateway/platforms/telegram.py: apply _effective_update_message to
  _handle_media_message (mirrors _handle_location_message)
• tests/gateway/test_telegram_channel_posts.py: add
  test_media_handler_uses_effective_message_for_channel_post; extend
  _make_channel_message with media attribute stubs

Test plan

• scripts/run_tests.sh tests/gateway/test_telegram_channel_posts.py -q → 4/4
• scripts/run_tests.sh tests/gateway/test_telegram_group_gating.py tests/gateway/test_telegram_channel_posts.py -q → 30/30

[alt-glitch]

Duplicate of #28614 — same fix applying _effective_update_message() to _handle_media_message() for channel_post media routing. Both extend merged #28531 which fixed text/commands but missed media.

[teknium1]

Thanks for the focused fix. I verified the premise against current remote main (733472952a0fb): gateway/platforms/telegram.py:5995 still returns early on not update.message, while the sibling text/command/location handlers already use _effective_update_message(update) at gateway/platforms/telegram.py:5780, 5797, and 5812.

Problems

• Current main has moved since this PR. _handle_media_message now also has an observe-unmentioned branch that still reads update.message at gateway/platforms/telegram.py:5997-6010. A salvage should normalize that branch to the same msg object too, not just the original early guard.

Suggested changes

• Apply the PR’s pattern on current main: compute msg = self._effective_update_message(update) once, then use msg throughout _handle_media_message.
• Keep the channel-post media regression test; the PR’s test targets the right behavior.

This is an automated hermes-sweeper review.

[teknium1]

Thanks for the focused regression fix. The premise remains valid on current main.

Problems

• The active adapter moved from gateway/platforms/telegram.py to plugins/platforms/telegram/adapter.py in 560010547, so this branch cannot be applied unchanged.
• Current _handle_media_message still rejects a channel post at plugins/platforms/telegram/adapter.py:7764, and its auth/observe paths continue to read update.message at :7766-7786. Resolving only the original guard would leave those branches inconsistent with the effective message.

Suggested changes

• Salvage the fix in plugins/platforms/telegram/adapter.py: bind msg = self._effective_update_message(update) once and use it throughout _handle_media_message.
• Port the media channel-post regression test into tests/gateway/test_telegram_channel_posts.py; current coverage there stops at text and command handlers (:150-181).

This is an automated hermes-sweeper review.

[teknium1]

The active adapter has since moved to plugins/platforms/telegram/adapter.py (5600105). When salvaging this line, use the resolved msg for the current handler's authorization and unmentioned-group observation branches too; they still read update.message directly at current main :7766-7786.