ci: avoid using pouchdb-server in testing, v2

[espy]

Destroying databases is flaky in pouchdb-server. This PR supersedes #9202

Overview

PouchDB integration tests were very flaky, with random timeouts. We (NH) identified this being caused by an issue with pouchdb-server, where the destroy() method used to clean up test DBs didn’t actually remove the test databases. The tests no longer seem to time out when using a CouchDB via COUCHD_HOST instead of pouchdb-server. This PR changes the CI setup so we can verify this assumption by running the CI repeatedly.

We moved the node PERF tests out into their own nodejs-perf Job , since our changes conflicted with that particular bit of the node job.

We’re also trying out randomly selecting Ubuntu mirrors other than the terribly slow Azure one, see actions/runner-images#7048 (comment)

Testing recommendations

We want to re-run the test suite multiple times to se whether the timeouts are resolved, and if so, submit this as a fix.

Checklist

• I am not a bot
• This is my own work, I did not use AI, LLM's or similar technology for code or docs generation
• Code is written and works correctly
• Changes are covered by tests
• Documentation changes were made in the docs folder

[espy]

Well, turns out opening a new PR on the same branch does not jolt Github into doing the correct thing in the CI UI 🙄

Will make one more attempt with a new branch.