tests: handle accepted regressions in release builds

[nrainer-materialize]

A known performance and scalability regression was introduced in v0.80.0. While regular nightly builds did not fail because because the regressions were marked as accepted, this was not the case for the release build.
This PR considers known regressions when running nightly builds for releases, which are based on versions.

Commits

cfce28b docker: add utility methods to classify image tag
6ceb50c tests: identify accepted regressions between two versions
89cafdb scalability framework: refactorings: extract methods
244a70a scalability framework: store resolved target of endpoints
bf2fa67 scalability framework: collect endpoints with regressions
1d8b4c5 scalability framework: print assessment, do not fail on justified regressions
c2f222c feature benchmark: refactoring: variable usages
4cbfcd7 feature benchmark: print assessment, do not fail on justified regressions

Status

Draft, need to go over the code myself again. ✅
Please pay special attention to c2f222c, which seems risky.

Nightlies

• Nightly of this branch + fixed version for ancestor and current: builds

[nrainer-materialize]

Nightly comparing v0.80.0 with v0.79.1 as baseline worked as expected:
✅ Builds succeeded
✅ Scalability test prints:

+++ Assessment of regressions
* Although there are regressions between baseline v0.79.1 (38a21b1a6) - release: bump to version v0.79.1 and endpoint v0.80.0 (15e58e51d) - release: bump to version v0.80.0, they can be explained by the following commits that are marked as accepted regressions: c4f520a57a3046e5074939d2ea345d1c72be7079.

✅ Feature benchmark prints:

ERROR: The following scenarios have regressions: InsertMultiRow, ConnectionLatency, SubscribeParallelTableWithIndex, StartupLoaded, SubscribeParallelKafka
However, the regressions are accepted. Accepted regressions were introduced with these commits: ['c4f520a57a3046e5074939d2ea345d1c72be7079']