Implement stabilizer for (transitive?) G-sets

[fingolfin]

I would expect stabilizer(gset) to return the stabilizer of representative(gset) with respect to the action.

Moreover, I would expect it to cache that stabilizer.

This way it would be a very natural fit for e.g. representing conjugacy classes, or the result of orbit-stabilizer computations

This mirrors the GAP function StabilizerOfExternalSet

One might perhaps restrict this to transitive G-sets, though technically there is no reason as long as we are clear what it means.

See also issue #108

[fingolfin]

@ThomasBreuer @mjrodgers it would be great if you could resolve this feature request

[mjrodgers]

If the action is not transitive, then we could have different results depending on which representative is chosen. How should this be handled in this case?

[ThomasBreuer]

As Max wrote, representative(gset) is the reference point for stabilizer(gset).
Besides that, I think stabilizer(gset, pnt) can be supported, without caching.

(Jens Brandt @flyingapfopenguin had talked to me about this feature just an hour ago. I will provide a pull request.)

[mjrodgers]

Yes, I guess I understand that gset will have some order put on it in the background and then this will return stabilizer(gset, gset[1]) but I guess I am just uncomfortable that the representative could be different depending on various factors that could decide what the reference point is. But if this is consistent with GAP then I suppose there is no issue.

[lgoettgens]

I would already be happy with stabilizer(gset) only for transitive G-sets, and stabilizer(gset, pnt) for the general case.