t/parallel.t fails (with newest Test2?)

[eserte]

The test suite started to fail on most systems. See https://fast2-matrix.cpantesters.org/?dist=Test-Class-Moose%200.99 for an overview. Statistical analysis suggests that this happens with newer Test2 versions (1.302202 and newer):

****************************************************************
Regression 'mod:Test2'
****************************************************************
Name           	       Theta	      StdErr	 T-stat
[0='const']    	      1.0000	      0.0000	18826770914471884.00
[1='eq_1.302162']	      0.0000	      0.0000	   1.81
[2='eq_1.302164']	     -0.0000	      0.0000	  -1.48
[3='eq_1.302182']	      0.0000	      0.0000	   0.00
[4='eq_1.302183']	      0.0000	      0.0000	   0.00
[5='eq_1.302186']	     -0.0000	      0.0000	  -0.60
[6='eq_1.302187']	     -0.0000	      0.0000	  -1.81
[7='eq_1.302188']	      0.0000	      0.0000	   0.00
[8='eq_1.302190']	      0.0000	      0.0000	   1.98
[9='eq_1.302191']	     -0.0000	      0.0000	  -5.30
[10='eq_1.302195']	     -0.0000	      0.0000	  -2.41
[11='eq_1.302198']	     -0.0000	      0.0000	  -2.96
[12='eq_1.302202']	     -1.0000	      0.0000	-15371994081575846.00
[13='eq_1.302203']	     -1.0000	      0.0000	-17186411858064022.00
[14='eq_1.302204']	     -1.0000	      0.0000	-18324633198293608.00

R^2= 1.000, N= 218, K= 15
****************************************************************

[autarch]

Yeah, it seems like a change in a recent version of Test2 that makes it incompatible with something Test-Class-Moose is doing internally.

Unfortunately, TCM is pretty complicated and it's been a long time since I've looked at it closely. I really don't know what the right fix is here.

[exodist]

Nothing obvious is jumping out at me, and the release @eserte identified seems to just add a conditional comment to output that should not interfere with this kind of test.

I will try to make time to review this as soon as I can, but it may be a few days.