Fix stabilizer computation in alternating groups which could sometimes return a wrong result

[fingolfin]

Fixes #5808

[fingolfin]

@stertooy can you confirm this fixes the issue for you?

[hulpke]

Thank you -- indeed this has been an awkward oversight.

[stertooy]

The fix works perfectly, many thanks!

I do have one small suggestion for the test.

[stertooy]

Perhaps something along the lines of

gap> stab := Stabilizer( G, One(G), tc );;
gap> Set( stab ) = Set( Filtered( G, g -> tc( One( G ), g ) = One( G ) ) );
true

would be safer?

This would ensure the following case (which I only discovered just now, and which is fixed by this PR) is also caught:

gap> G := AlternatingGroup( 6 );;
gap> hom1 := GroupHomomorphismByImages( G, G, [ (1,2,3,4,5), (4,5,6) ], [ (1,2,6,3,5), (1,4,5) ] );;
gap> hom2 := IdentityMapping( G );;
gap> tc := function ( g, h ) return (h^hom2)^-1 * g * h^hom1; end;;
gap> stab := Stabilizer( G, One(G), tc );;
gap> Size(stab);
5
gap> Set( stab ) = Set( Filtered( G, g -> tc( One(G), g ) = One(G) ) );
false

[fingolfin]

I've added your second test case, and made the checks more robust