Fix Slack multi-workspace token and routing regressions

[kshitijk4poor]

Summary

• skip invalid cached Slack workspace tokens during startup instead of aborting the adapter
• persist channel_id -> team_id routing to slack_channel_teams.json so outbound sends after restart reach the correct workspace
• allow outbound Slack client selection to use explicit team_id metadata when available
• add focused Slack regression tests for stale saved tokens and persisted channel routing

Why

The upstream multi-workspace Slack change introduced two regressions:

1. one revoked token in slack_tokens.json could abort connect() and keep Slack fully offline
2. proactive sends to secondary workspaces after restart could fall back to the primary client and fail with channel_not_found

This PR makes token validation per-workspace and persists learned channel routing so workspace selection survives process restarts.

Validation

• source .venv/bin/activate && python -m pytest tests/gateway/test_slack.py -q
• source .venv/bin/activate && python -m py_compile gateway/platforms/slack.py tests/gateway/test_slack.py

[teknium1]

Closing as superseded by #70190 (merged): predates the plugin migration and most of its targets were already fixed on main; remaining gaps covered by #70190.

Thanks for the work — it's credited in #70190's summary.