Clean up tests to diagnose Frank-Wolfe linear solver error

[gdalle]

• Revert specification of test dependencies to the more robust version (see this issue)
• Adjust default parameters for Frank-Wolfe (increase precision and switch step size selection) because a recent release or FrankWolfe.jl broke our tests (see Adaptive line search for Frank-Wolfe breaks tests #40)
• Deactivate test verbosity to make CI logs easier to read
• Switch data to Float32 to suppress Flux.jl warning
• Try to parallelize tests but they no longer show up in the test set so I gave up (see this discourse thread, it seems like an open problem)

[codecov-commenter]

Codecov Report

Patch coverage: 25.00% and project coverage change: -0.27 ⚠️

Comparison is base (49dfd06) 89.26% compared to head (a840f89) 89.00%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #39      +/-   ##
==========================================
- Coverage   89.26%   89.00%   -0.27%     
==========================================
  Files          18       18              
  Lines         382      382              
==========================================
- Hits          341      340       -1     
- Misses         41       42       +1     

Impacted Files	Coverage Δ
src/frank_wolfe/frank_wolfe_utils.jl	83.33% <ø> (ø)
src/frank_wolfe/differentiable_frank_wolfe.jl	93.75% <25.00%> (-6.25%)	⬇️

... and 2 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

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

[gdalle]

@BatyLeo maybe switching to agnostic step size is not a good idea, see ZIB-IOL/FrankWolfe.jl#387 (comment)

Let's wait a bit before we merge

[BatyLeo]

@BatyLeo maybe switching to agnostic step size is not a good idea, see ZIB-IOL/FrankWolfe.jl#387 (comment)

Let's wait a bit before we merge

Ok 👍