Avoid using structure() in tests

[MichaelChirico]

Follow-up to #7813.

structure() is slow, and, as we saw there, can be a bit fragile. It's better to use the greenlit path to construct objects where possible.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.01%. Comparing base (f55ed35) to head (bb021b4).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7814   +/-   ##
=======================================
  Coverage   99.01%   99.01%           
=======================================
  Files          88       88           
  Lines       17228    17234    +6     
=======================================
+ Hits        17059    17065    +6     
  Misses        169      169           

☔ View full report in Codecov by Harness.
📢 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.

[github-actions]

• HEAD=structure faster P<0.001 for DT[by,verbose=TRUE] improved in #6296
  

Generated via commit bb021b4

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

Task	Duration
R setup and installing dependencies	3 minutes and 3 seconds
Installing different package versions	23 seconds
Running and plotting the test cases	5 minutes and 33 seconds

[ben-schwen]

Overall LGTM

When setting class() and subsequently setting attr() we could also set both with attributes()

[MichaelChirico]

When setting class() and subsequently setting attr() we could also set both with attributes()

Not quite so simple, e.g. for building xt in other.Rraw, since attributes<- also wipes dim/dimnames/names. mostattributes<- might be appropriate, but the current approach looks ~OK.