Add accessors group and subgroup for SubgroupTransversal

[mjrodgers]

We add the following accessors for T::SubgroupTransversal a (left or right) transversal of a subgroup H in G:

• group(T) for T.G
• acting_group(T) subgroup(T) for T.H
  along with docstrings that are easy to find in the main documentation.

These accessor names are consistent with those for a GroupCoset.

Closes #4296

[fingolfin]

But in how far is this group "acting"? If we look at the (right) cosets in $G/H = { Ha \mid a \in G }$ then I'd say this is a $G$-set, not an $H$-set (well, since $H$ is a subgroup, of course it would be valid to "restrict it" to $H$, but the most general starting point is to view it as a $G$-set.

Of course now we talk about a "transversal", which in general is neither a $G$-set nor an $H$-set. But one can in principle try to "fix" this, as follows: for `$r\in T$ and $g\in G$, define $r.g$ as the element $r' \in T$ for which $Hrg = Hr'$ (and at that point $rg/r'$ is a Schreier generator of $H$). This again suggests $G$ as acting group. Don't get me wrong: I am not saying we should implement such an action. Rather, I am trying to find with a rational for this naming, and coming up empty.

What am I missing?

[mjrodgers]

hm, I see your point. I wanted to use the same method names as for the individual cosets; but I agree, in this context G is the group that would be acting on the cosets.

I'm not really sure what would be a well-motivated way to name these methods (Magma doesn't have anything analogous, since they just treat a transversal as a generic ordered set)

[fingolfin]

@ThomasBreuer and me will think about it and try to make a suggestion

[ThomasBreuer]

I would say G == group(T) and H == subgroup(T) are suitable, if there is really need for such functions.

(Technically, one needs G in order to set it as the parent of an element when one asks for T[i], but this is only for internal use. Logically, a transversal is just a list of group elements, and the same transversal can be used for different subgroups. In this sense, fetching H from a transversal T is a bit artificial.)

[mjrodgers]

Ok, this has been renamed - as long as all tests still pass, it should be ready to merge.

[fingolfin]

Doctests fail because subgroup is not exported

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.89%. Comparing base (4ea3c46) to head (9d3c70b).
Report is 50 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4839      +/-   ##
==========================================
+ Coverage   84.74%   84.89%   +0.14%     
==========================================
  Files         683      683              
  Lines       91579    91942     +363     
==========================================
+ Hits        77613    78053     +440     
+ Misses      13966    13889      -77     

Files with missing lines	Coverage Δ
src/Groups/cosets.jl	95.81% <100.00%> (+0.04%)	⬆️

... and 81 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.