Improve compatibility with GAP's MatrixObj project

[fingolfin]

... by using TryNextMethod in Matrix() methods instead of raising an error

Together with gap-system/gap#4833 by @hulpke this should resolve the conflict between GAP master and semigroups. So if we this patch is acceptable for you, then as soon as there is a semigroups release with it, we could merge that GAP PR and finally release GAP 4.12 -- well, after some more testing, obviously! But thanks to the work done on https://github.com/gap-system/PackageDistro we will also (again) get tests of Semigroups (and all other distributed GAP packages) against GAP master, so we can deal with those, too

[james-d-mitchell]

I’m happy with this, does it really fix the issues with master? If so, then this seems pretty straightforward and I’ll be happy to merge if you could rewrite the tests ( or I can do this), thanks for the pr @fingolfin

[fingolfin]

In my tests, it fixed the issue together with gap-system/gap#4833 (which is not yet merged as it would break our CI). I am not saying this resolve all potential problems, but at least the one at hand.

I am at a workshop this week, so I am not sure when I'll be able to look into fixing the tests. If you can do it quickly, that'd be great, otherwise I'll try it once I am free again.

[fingolfin]

I've discovered a few more places that need TryNextMethod, and adjusted some of the tests.

But I think I'll also have to disable a few tests, simply because the behaviour in 4.10 / 4.11 will be different from that in 4.12. And arguably some bugs ought to be fixed in GAP master, too... but one step after the other...

[fingolfin]

@james-d-mitchell I've updated this again (sorry for taking so long). Assuming this PR passes tests, and you are OK with it, I think we'd have to proceed as follows:

1. merge this PR
2. release semigroups 4.0.1 with it
3. merge that update into the GAP package distro (easy :-) )
4. merge Remove upranking of method that collides with semigroups package. gap-system/gap#4833
5. once all that works, we can remove some of the "workarounds" in this PR again (they mostly exist to solve the chicken-and-egg problem between merging this PR first vs. merging that GAP PR first)

[james-d-mitchell]

Thanks @fingolfin I'm happy to merge this, and with the rest of the steps too. I'm a wee bit confused why there are 3 commits, when 2 of them are in master already. I tried rebasing and pushing to your branch but it says it's up to date?

[fingolfin]

Oh, sorry: this PR is targeting master, but perhaps it should be targeting stable-4.0? I can change that, let me know

[james-d-mitchell]

Either branch should be ok, maybe rebasing on the present master will be sufficient?

[fingolfin]

Changing the base branch is possible with a button press from my phone, so I did that now :-)