Optimize ci by reusing runners

[jimtng]

Resolve #310

• I started off from @boc-tothefuture's work.
• a @log_level_changed was missing in one of the scenarios in rule_language.feature. The scenario set the log level to INFO causing subsequent tests using trace log to fail
• added index to the matrix array and use it in openhab dump file. Previously it used the very long feature names which also contain illegal characters : and /

Other non-essential changes:

• Added --no-heap-dump to dev:dump-create to reduce the openhab dump file size (in the case of test failures)
• Updated test against 3.3.0.M5
• Replaced / simplified several ruby setup actions with ruby/setup-ruby + bundler-cache in workflow.yml
• Close the file handle in truncate_log - I'm not sure whether the file handles would've gotten closed anyway by Ruby's GC at the end of the method call, but it's better to be sure

This brings the test time down from 20m to 12m - 15m

I believe this should be ready to be merged

[jimtng]

Note: cucumber --retry in test.rake doesn't seem to be doing what it's supposed to do. What I believe we want it to do:

• if it failed, retry, and if the second attempt is successful, continue to the next test

• if subsequent retry is successful (up to the specified limit 3), then consider the test successful, and exit with error code 0

  What it actually does:

  • if it failed, retry 3 times, even when all the subsequent attempts were successful, still do it 3 times in addition to the initial failure, making it run 4 times.
  • Even when subsequent attempts were successful, exit with exit code 1, causing the entire thing to fail.

In this case, cucumber flags the test as flaky hence it returned the exit code 1

[github-actions[bot]]

:tada: This PR is included in version 4.42.0 :tada:

The release is available on:

• 4.42.0
• GitHub release

Your semantic-release bot :package::rocket: