Temporary PR to evaluate GitHub workflow output

[jppelteret]

No description provided.

[jppelteret]

@Rombur FYI. I put together this branch to evaluate and fix the CI test failures. I took an educated guess as to how to fix some of the remaining issues, but it seems that those changes have not been 100% effective. There are still 4 failing tests after my last commit

	 96 - weak_forms/step-44-variant_02-residual_view_ad_02b.debug (Failed)
	141 - weak_forms/step-72-variant_01b_vec.debug (Failed)
	142 - weak_forms/step-72-variant_01c.debug (Failed)
	168 - weak_forms/step-transient_curl_curl-variant_02b.mpirun=3.debug (Failed)

Since I cannot reproduce these on my Mac, I've built a docker image on a Windows machine and am busy running all of the tests through that. Hopefully that will allow me to debug the at least 3 of the 4 failures. The step-72 tests are giving a floating point exception (not treated by the changes I made to the division binary operator), so I need to debug those using GDB. Even though I changed the evaluation point for the step-transient_curl_curl test, one still fails for some reason. Maybe I'll just extract some alternative output, as I need to refactor the test anyway (the mesh needs to change, amongst other things). I fear that the step-44-variant_02-residual_view_ad_02b test might not be treatable on the CI unless we build ADOL-C ourselves. I guess that the ubuntu package does not have the necessary build flags set (specifically the --enable-traceless-refcounting flag), but I do still need to double check this.

[jppelteret]

Superseded by #43 in conjunction with #50 and #58