tests: import: check that dvcfs is using main repo's cache

[efiop]

Followup for #9415

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: -0.25 ⚠️

Comparison is base (967dd5e) 91.59% compared to head (6c98eaf) 91.34%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #9417      +/-   ##
==========================================
- Coverage   91.59%   91.34%   -0.25%     
==========================================
  Files         487      487              
  Lines       37779    37780       +1     
  Branches     5436     5436              
==========================================
- Hits        34602    34511      -91     
- Misses       2620     2693      +73     
- Partials      557      576      +19     

Impacted Files	Coverage Δ
tests/func/test_import.py	99.46% <100.00%> (+<0.01%)	⬆️

... and 23 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[shcheklein]

Thanks, Ruslan. Assuming the "by design" behaviour of *FS is tested in their corresponding tests, it's good and should be enough. (I would personally still sleep better with some integration tests that cover such things).

[efiop]

@shcheklein I've thought about a more comprehensive test, but couldn't come up with something comprehensive but also quick, not fragile and reasonably simple. Might come back in the future. Same with tweaking the current benchmark. I will keep it in mind.

[skshetry]

I think this is dvcfs/datafs design issue, as there's always a default cache. If it did support passing an odb, we would not have to pass the config. Also, passing concrete types are always better than passing config dicts.

https://github.com/iterative/dvc/blob/6b6084a84829ba90b740a1cb18ed897d96fcc3af/dvc/fs/dvc.py#L260-L261

We would not have to do things like this for plots if it had a notion of default cache.