Generic overapproximate by falling back to convert

[schillic]

Instead of duplicating these overapproximate(X, Y) = convert(Y, X) methods, we could define one such fallback method that checks for hasmethod and otherwise throws an error.

[mforets]

Yes, i thought about this recently. I think it is a good idea, but it may need to fix ambiguities for cases that are already generic in the set type, e.g. for overapproximate(::LazySet, Hyperrectangle), we would not like to use the convert fallback (always).

[schillic]

True, but I hope that Julia will consider the other methods as "more specific" and choose them automatically.