Fix RankOfPartialPermSemigroup for groups

[wilfwilson]

Previously, RankOfPartialPermSemigroup for a group of partial permutations would give an unexpected error if the GeneratorsOfGroup were empty (it delegated to RankOfPartialPermCollection(GeneratorsOfGroup(G))). A simpler things to do for groups, and for monoids actually, is to call RankOfPartialPerm(One(G)).

Resolves #3037.

[olexandr-konovalov]

@wilfwilson are you sure that this had to be backported? I did it, and now with Semigroups package (version 3.0.20) loaded I observe this in the stable-4.10 branch:

########> Diff in /circa/scratch/gap-jenkins/workspace/GAP-minor-release-test/\
GAPCOPTS/64build/GAPTARGET/install/label/kovacs/gap-4.10.1/tst/testinstall/sem\
ipperm.tst:90
# Input is:
RankOfPartialPermSemigroup(S);
# Expected output:
2
# But found:
Error, no method found! For debugging hints type ?Recovery from NoMethodFound
Error, no 1st choice method found for `RankOfPartialPermCollection' on 1 argum\
\
ents
########

[wilfwilson]

@alex-konovalov Yes i wanted it to be backported to stable-4.10, although it didn’t have to be. However, I’m sorry, but I did not test whether the Semigroups package ran properly when this was backported, and unfortunately this causes a test failure, as you have found.

There are two reasons for this: in stable-4.10 the method ordering is still incorrect (this is largely/completely fixed in master now, thankfully). Because of this, for the newly added test in this PR, the incorrect method for DegreeOfPartialPermSemigroup is being applied. This is unfortunate, but the other, less desirable, applicable methods should still work. Secondly, the method that is being called instead causes an unexpected error, because of a newly discovered bug in the Semigroups package. So, this is caused by the method reordering problems in GAP and a bug in Semigroups, in combination.

I have fixed the bug in Semigroups here: semigroups/Semigroups#565 and so this problem will go away once that is merged and the next version of Semigroups is released.

If you are impatient for this test failure to be resolved, you could revert the backport to stable-4.10: I can live without this PR being in stable-4.10.

[olexandr-konovalov]

Thanks - if it's a matter of week until @james-d-mitchell will be making a new release of Semigroups, I'd rather wait till revert.

[olexandr-konovalov]

Backporting of this change to stable-4.10 branch was reverted in 465d77e because we need to wait for the Semigroups package update first.