Improve FastNormalClosure

[ssiccha]

Add the RecogNode as argument.

Drop the undocumented assumption, that the group generated by is an overgroup of the one generated by . Thus change a test whether the former group is cyclic into a test whether the entries of

commute with the ones in .

[codecov[bot]]

Codecov Report

Merging #304 (c0c6aea) into master (4842ec6) will increase coverage by 0.48%. The diff coverage is 94.11%.

@@            Coverage Diff             @@
##           master     #304      +/-   ##
==========================================
+ Coverage   78.13%   78.61%   +0.48%     
==========================================
  Files          43       43              
  Lines       18427    18433       +6     
==========================================
+ Hits        14398    14492      +94     
+ Misses       4029     3941      -88     

Impacted Files	Coverage Δ
gap/projective/d247.gi	75.23% <50.00%> (ø)
gap/generic/kernel.gi	90.08% <100.00%> (+0.51%)	:arrow_up:
gap/projective/almostsimple.gi	66.04% <100.00%> (ø)
gap/projective/classicalnatural.gi	94.10% <100.00%> (ø)
gap/projective/findnormal.gi	78.17% <100.00%> (ø)
gap/projective/tensor.gi	55.86% <100.00%> (ø)
gap/projective/c6.gi	62.17% <0.00%> (+0.18%)	:arrow_up:
gap/perm/giant.gi	93.04% <0.00%> (+2.46%)	:arrow_up:
gap/projective/AnSnOnFDPM.gi	89.63% <0.00%> (+9.60%)	:arrow_up:

[ssiccha]

I could combine this with the issue in #288. But in some situations the generators of the group which are taken do not come with a RecogNode. I'll have to think about this again.

[ssiccha]

I got rid of the whole recog node shenanigans. This is ready for review now.