fix(slack): throttle repeated channel directory warnings

[LeonSGP43]

Summary

• Throttle repeated Slack channel-directory refresh warnings by team/error so recurring Slack API failures do not flood gateway logs.
• Keep the first warning visible and emit suppressed repeats at debug level.
• Add regression coverage for repeated workspace failures.

Fixes #18485
Signal Deck card: ops-ops-channel-directory-failed-to-list-slack-channels-for-team-the-request-to-the-slack-api-

Tests

• scripts/run_tests.sh tests/gateway/test_channel_directory.py::TestBuildSlack::test_repeated_workspace_errors_are_warning_throttled tests/gateway/test_channel_directory.py::TestBuildSlack::test_per_workspace_error_does_not_block_others tests/gateway/test_channel_directory.py::TestBuildSlack::test_response_not_ok_breaks_pagination_for_that_workspace
• scripts/run_tests.sh tests/gateway/test_channel_directory.py
• git diff --check origin/main...HEAD

[teknium1]

Thanks for the focused Slack log-noise fix. Current main still logs each users.conversations non-OK response at gateway/channel_directory.py:243 and each workspace-client exception at gateway/channel_directory.py:264; gateway housekeeping refreshes the directory every five minutes (gateway/run.py:20123-20133). The PR routes both paths through the same team/error-keyed monotonic throttle and adds regression coverage for repeated non-OK workspace responses.

No correctness problems were verified in the proposed change. This is an automated hermes-sweeper review.

[teknium1]

Merged via #70192 — your commit was cherry-picked/reapplied onto current main with your authorship preserved in git history: your channel-directory warning throttle was cherry-picked.

Thanks for the contribution!