Refactor cedta() for readability

[MichaelChirico]

As the set of checks here grows, the serial || approach is getting harder-and-harder to parse.

Instead, we split these tests out into a helper.

[github-actions]

• HEAD=cedta-sd slower P<0.001 for setDT improved in #5427
• HEAD=cedta-sd slower P<0.001 for memrecycle regression fixed in #5463
  

Generated via commit bc1d326

Download link for the artifact containing the test results: ↓ atime-results.zip

Task	Duration
R setup and installing dependencies	2 minutes and 58 seconds
Installing different package versions	40 seconds
Running and plotting the test cases	2 minutes and 38 seconds

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.77%. Comparing base (67670e9) to head (bc1d326).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #7162      +/-   ##
==========================================
- Coverage   98.77%   98.77%   -0.01%     
==========================================
  Files          81       81              
  Lines       15185    15182       -3     
==========================================
- Hits        14999    14996       -3     
  Misses        186      186              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jangorecki]

This looks like a risky change now when approaching major release. What do you think about merging it just after 1.18.0 release, so it has more time to mature sitting in devel.

[MichaelChirico]

you think the refactoring part is risky, or the new rule? possibly we could tease them apart. I am fine either way, this is just bug triage, not particularly crucial.

(btw, I'm not sure how close we actually are to release)

[jangorecki]

Just a heavy rework of the internal mechanism that packages depends on... I haven't look into the changes much but doing such changes before release is like asking for troubles.

[MichaelChirico]

IMO the change is pretty simple, low-risk -- just changing a bunch of || checks into individual early returns inside a helper. And any possible issue would very likely be caught by revdep checks since [ must be called O(1M) times in downstreams.

Anyway, I've separated the refactor from the actual bug-fix: #7166.