Remove special handling of stdlibs in test_deps_compat

[lgoettgens]

Resolves #212.

[codecov]

Codecov Report

Merging #215 (b662e48) into master (69a7655) will decrease coverage by 0.93%.
Report is 1 commits behind head on master.
The diff coverage is n/a.

❗ Current head b662e48 differs from pull request most recent head 40176e2. Consider uploading reports for the commit 40176e2 to get more accurate results

@@            Coverage Diff             @@
##           master     #215      +/-   ##
==========================================
- Coverage   89.63%   88.71%   -0.93%     
==========================================
  Files          10       10              
  Lines         415      505      +90     
==========================================
+ Hits          372      448      +76     
- Misses         43       57      +14     

Flag	Coverage Δ
unittests	88.71% <ø> (-0.93%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
src/deps_compat.jl	70.27% <ø> (-8.31%)	⬇️
src/utils.jl	89.85% <ø> (+4.60%)	⬆️

... and 8 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[vtjnash]

This (or something surrounding it) seems to have caused quite a few failures in PkgEval, which reduces the overall testing coverage in the ecosystem. Is there something we can do about this? For example:
https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_hash/1fc9d54_vs_72cd63c/Quadmath.primary.log
https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_hash/1fc9d54_vs_72cd63c/AbstractTrees.primary.log
https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_hash/1fc9d54_vs_72cd63c/ManualMemory.primary.log

Looking at the apparent red cliff in the report: https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_hash/1fc9d54_vs_72cd63c/report.html

Some don't seem precisely related, as the error appears to complain about packages (e.g. BenchmarkTools) that don't appear to even be installed in the log?

[lgoettgens]

Yes, parts of the failures you linked are due to this change. But a lot are due to checking extras and weakdeps in general since #202. So every package declaring Aqua as a extra-Dep without a compat bound will fail as well.

The clean solution would be to change all failing packages to either fix the issues there or (easier) add a compat entry for Aqua 0.7 there.
If you are interested in PkgEval in particular, I can think about changing some kwargs in the default test list depending on the PkgEval ENV variable, so by default PkgEval testing would only use a subset of Aqua tests. Would that be something you are interested in? If the list of failing packages is rather short, I think this is not worth the effort and we should instead provide PRs to them. I unfortunately don't really feel confident to condense such a list from the PkgEval report you linked above.