Remove orphaned deploy workflows and harden API replacement

[ComputelessComputer]

Summary

Problem: Removed bot and internal Slack services still had CI or deployment workflows, while API rolling replacement could fail when Fly had not propagated a newly pushed image manifest yet.

Fix: Delete the orphaned workflows and retry only Fly manifest-propagation errors during replacement-machine creation, with a focused regression test.

Verification

• GitHub checks on this draft PR

---

Note

Low Risk
Workflow deletions only affect removed apps; deploy retry is narrowly scoped to manifest propagation errors during machine creation.

Overview
Removes orphaned GitHub Actions for services that no longer ship: bot_ci.yaml, bot_cd.yaml, and slack_internal_cd.yaml (bot typecheck/test and Fly deploys for bot and slack-internal).

Hardens API drain deploy by retrying replacement-machine creation when Fly returns HTTP 400 for a not-yet-available image manifest (MANIFEST_UNKNOWN / manifest unknown), using the same attempt count and exponential backoff as other Fly API retries in deploy_api_drain.py. A regression test asserts a failed first POST is retried after 0.5s and then succeeds.

^{Reviewed by Cursor Bugbot for commit 4dfde62. Bugbot is set up for automated code reviews on this repo. Configure here.}

[netlify]

✅ Deploy Preview for anarlog canceled.

Name	Link
🔨 Latest commit	4dfde62
🔍 Latest deploy log	https://app.netlify.com/projects/anarlog/deploys/6a96c7ab07312e00087fb01d