Make RestrictedMapping of a group homomorphism use the same Range as the original morphism

[stertooy]

Closes #6225

Text for release notes

see title

Further details

Shouldn't have much impact, unless some other code actually relied on RestrictedMapping returning a surjective map.

[ThomasBreuer]

The proposed changes are correct.

In another pull request, we will eventually

• extend the documentation of RestrictedMapping in order to state what the return value is,
• make sure that all methods of RestrictedMapping fit to this description,
• document the surjective option that is used already in at least one of these methods.

[fingolfin]

Why is this removed? If the map was total (= not partial) then surely its restriction is also total? And this is kinda important to know about a map

[stertooy]

IsGroupHomomorphism implies IsTotal, so both hom and rest should have IsTotal set to true already.

[fingolfin]

I agree that the range should not be touched (although it would be useful, nay important, to have a function doing that as well, and ideally do both in one go). However, I am wary this may break packages, or user code, so I've started a run of the package distro test suite. Of course even if it fails, that doesn't mean the end of the PR, but then we need to carefully evaluate the failures.

And if it passes: great!

BTW a long time tree was to refactor the whole mapping code, separating the part of the map which does the actual computation (the ... "underlying function" ? hrm, don't have a good word for that... basically, the $\mapsto$ part) out, so that restrictions and co-restrictions are essentially free (they can all share the same/identical "mapsto-part")