regression in GAP 4.13.0, concerning IsomorphismPermGroupOrFailFpGroup?

[ThomasBreuer]

Consider the following input.

F:= FreeGroup(2);
gens:= GeneratorsOfGroup( F );
x:= gens[1];; y:= gens[2];;
rels:= [ y*x^-1*y^-1*x*y^-1*x^-1, x^-1*y*x*y*x^-1*y^-2 ];;
G:= F / rels;
IsomorphismPermGroupOrFailFpGroup(G, 100000);

In GAP 4.12.2, this yields an isomorphism.
In GAP 4.13.0 and in the current master branch, I get fail.

[fingolfin]

This regression was introduced by PR #5229 resp. commit df91fdb (ping @hulpke)

[fingolfin]

Thomas' group is just SL(2,5) in disguise:

gap> Size(G);
120
gap> StructureDescription(G);
"SL(2,5)"

[fingolfin]

Ah OK, so the coset limit is highly relevant. It works with the default CosetTableDefaultMaxLimit$= 4096000$ but not with Thomas' smaller limit. Nor with CosetTableDefaultMaxLimit/10 it turns out. So 409600 cosets are not enough for this group of size 120?

[ThomasBreuer]

To be precise: The Size attribute of G is already set when IsomorphismPermGroupOrFailFpGroup returns fail. How can this happen?