keep or remove RegionMixin

[jGaboardi]

Based on @sjsrey's comment, we should a more thoughtful discussion on whether to keep or remove RegionMixin. It does not hurt to keep it around until we are sure we want/need to remove it.

[jGaboardi]

xref #294

[jGaboardi]

So here's a overview of where we're at on 'base'-level classes:

• spopt/BaseClass.py
  • should probably be renamed to base.py
  • houses BaseSpOptSolver, BaseSpOptExactSolver(BaseSpOptSolver), and BaseSpOptHeuristicSolver(BaseSpOptSolver)
• spopt/region/base.py
  • houses RegionMixin(object)
• spopt/locate/base.py
  • houses LocateSolver(BaseSpOptSolver)
  • and other Mixin-type classes for the various locate models

So it seems that RegionMixin can still hold value by being consumed in the individual regionalization classes. Consider maxp.py, for example:

• the MaxPHeuristic(BaseSpOptHeuristicSolver) class has a .solve() method the solves the heuristic MaxP problem then assigns labels and the values of p.
• the RegionMixin class has a .solve_assign() class.
• so we'll need to work out how to integrate these...

Also, within locate we are instantiating LocateSolver from BaseSpOptSolver, not BaseSpOptExactSolver, due to ortools still being used there for some reason (xref #18, #49, #61).

I guess I'm still just a bit confused about what we are gaining from having (up to) 3 levels of inheritance when models are still being solved from a method defined within a top-level class (e.g. MaxPHeuristic.solve()). Very willing to have a meeting to hash this out further because I'm sure it is something that I am just not understanding properly.

@sjsrey @knaaptime @gegen07 @ljwolf