Resolve AbstractAlgebra.FPModule vs Oscar.ModuleFP

[fingolfin]

Taken from #3054 where it is mentioned but without details.

Today I was asked which of the two to use for implementing a new module type (in this case: root spaces).

At the very least, we should rename one, or perhaps both, because the current very similar names just lead to confusion.

IIRC AbstractAlgebra.FPModule is dense and Oscar.ModuleFP sparse (as much as one can claim such a thing for abstract types which both are -- but FPModule has a lot of "generic" code which clearly assumes something dense.

[fingolfin]

Related to this, consider this from Slack. Someone asked how to make something similar to the following code work:

M1 = free_module(ZZ, 3);
M2 = free_module(QQ, 3);
f = hom(M1, M2, gens(M2)])

does not work.

As a workaround, this was suggested by @thofma (thank you!)

M1 = FreeMod(ZZ, 3);
M2 = FreeMod(QQ, 3);
f = hom(M1, M2, gens(M2)])

While I am glad we have a workaround, clearly this is really confusing. What's the difference between free_module and FreeMod? Why do both exist? How would one find FreeMod? etc.

I think this is tightly related to FPModule vs ModuleFP, hence added it here.