Clarify state of combinations

[fingolfin]

Today there was some confusion about combinations because clearly Oscar code uses it but Oscar.combinations does not exist.

Turns out AbstractAlgebra.combinations exists but is not exported.

Moreover we have Oscar.LieAlgebras.combinations which wraps the AA methods but calls sort on the result.

It would be nice to have this function (and perhaps also Oscar.LieAlgebras.multicombinations) as part of our Oscar combinatorics "package".

But I am guessing there is a reason why this is not just exported right now... perhaps someone (@thofma ? @lgoettgens ?) knows and can fill us in what has to be done.

(Once we know what is needed, perhaps @mjrodgers can take care of this)

[thofma]

The original version was added here Nemocas/AbstractAlgebra.jl#123 as a means to get the minors of a matrix.

The reason we do not export is that people have strong opinions about what the "right" order is, see the sort things in the LieAlgebras code base.

There is also Oscar.OrderedMultiIndexSet, which is yet another implementation of combinations, with some emphasis of iterating in the "right" order.

[fingolfin]

So how do we resolve this? Have an alg or ordering or so argument which allows to select an ordering, with some default the PIs select by vote/dice/whatever?