Return enum elements for overlap, and adapt ITF tests

[lbenet]

No description provided.

[lbenet]

After today's meeting I've checked EnumX and it looks pretty cool.

• Considering adding it as a new dependence seems ok to me, in the sense that it is a pretty light package.
• Using it would allow to export only the typename (Overlap is perhaps better than OverlapType), which brings along the different (enum) instances.
• We could get rid of the odd naming I proposed initially, e.g. _contains or _first_empty, in favor of Overlap.contains or Overlap.firstEmpty (as you suggest) or Overlap.first_empty.

@lucaferranti What's you opinion on giving it a try?

[lucaferranti]

@lbenet sorry for the delay in answering, I didn't notice the comment, I think we could give it a try, but does it have a significant advantage wrt returning a symbol?

[lucaferranti]

well we can try, it's not a biggie :)

[lbenet]

I tested using EnumX.jl: it corresponds to what we have in this PR, with a nicer interface, in the sense that only Overlap (the struct containing the overlap instances) is exported, instead of all specific instances. Shall I push the commit here so you can have a look and test it?

[lucaferranti]

@lbenet sure!

[lbenet]

I think this is ready @lucaferranti. Shall we backport it in an adapted form to master?

[lucaferranti]

@lbenet LGTM, merging. don't have strong opinions on whether or not we should backport this to master and release a new patch release. If 1.0-dev is going to be merged "soon", then I think we can wait.