Refactor overlap checking code

[pshriwise]

Right now the overlap check capability in overlap.cpp is linked into the XDG library itself, but until we're using that as part of the core library functionality we should reorganize the repo so that the code is in the tools directory. As it stands we currently need the indicators target to build XDG, which doesn't make sense when XDG_ENABLE_TOOLS is off.

Also, if we are using that capability in the core library down the lines, then we probably want to factor out indicators altogether.

[Waqar-ukaea]

So I have had some thoughts about this and what it could look like as we refactor the overlap checking code to be a part of the main XDG library, that sets us up nicely should we want to make use of overlap checking during particle tracking.

I am thinking that we expose some new method that lives in the XDG class like xdg::find_overlaps(array_of_vols, options, callback_hook_for_progress_bar). The main change I think I would like to see is something where the method can be called over a subset of volumes rather than always being over the whole model. Maybe a user just wants to check for overlaps between two volumes they know are problematic. This way they have the option to do so but can also do a more general check over the whole model still by passing the array of all volumes (minus implicit complement).

A couple of other notes:

• I'm thinking we keep the current setup where edge checking is an optional flag since it is more expensive than vertex checks
• We still keep an overlap-check miniapp but now it basically just acts as an example implementation of the API with some added UI:
  • The full verbose reporting move entirely to the realm of the miniapp and have the API just return the structured data?
  • Progress bars/reporting inserted via callbacks defined in the miniapp?
  • This way, we also still keep indicators as a dependency out of the core library
• Probably split up the somewhat monolithic check_instance_for_overlaps() into two explicit functions which pass over the vertices and edges independently - maybe still keep find_overlaps() as a thin wrapper which just calls the two separate methods?

[Waqar-ukaea]

I've re-opened this one because I think there is probably still some further work to do arguably