fix(feishu): preserve threaded replies

[stevenchouai]

Summary

• preserve Feishu/Lark root message ids as gateway thread routing targets
• fall back to metadata.thread_id when sending Feishu replies without an explicit reply_to
• keep streamed progress/status messages in the originating Feishu thread
• add regression coverage for inbound root_id handling and metadata-based threaded replies

Test Plan

• python -m py_compile gateway/run.py gateway/platforms/feishu.py
• python -m pytest tests/gateway/test_feishu.py -q -o 'addopts='

[stevenchouai]

Local validation update: GitHub reports no checks on this branch, so I ran the focused Feishu gateway suite locally with the repo wrapper. Result: scripts/run_tests.sh tests/gateway/test_feishu.py → 200 passed, 8 dependency deprecation warnings.

[stevenchouai]

Rechecked this before merge:

• python -m py_compile gateway/run.py gateway/platforms/feishu.py
• scripts/run_tests.sh tests/gateway/test_feishu.py -q

Result: 200 passed, 8 warnings locally. GitHub reports no checks for this branch. I tried to merge after validation, but this fork account does not have MergePullRequest permission on NousResearch/hermes-agent, so this still needs a maintainer merge.

[teknium1]

Thanks for the focused Feishu regression work. This is already implemented on current main.

• Automated hermes-sweeper review verified inbound root/thread routing in plugins/platforms/feishu/adapter.py:3252.
• Progress delivery uses the originating event as reply_to in gateway/run.py:17358, while status and stream paths carry reply_to_message_id in gateway/run.py:17800 and gateway/run.py:17987.
• Metadata-based threaded replies are implemented in plugins/platforms/feishu/adapter.py:4609 and covered by tests/gateway/test_feishu.py:2080.
• The equivalent salvage merged as fix(feishu): keep topic replies in threads (salvage #19750) #20851 (441ef75d157d6308a9f14d42a7b0ec8566866ef8; follow-up merge commit 28299afc21a37784d93b90924317f004ea2298af) and shipped in v2026.5.7.