Generic indicator names

[Joao-Dionisio]

Indicator constraints all had the same name by default. They now have the same default naming convention as other constraints.

I did discover a big bug that I'll fix later. When a model has indicator constraints and is optimized, all the constraints disappear. Not good. The tests are failing because of the assertion I added detecting this. EDIT: I'm very dumb, by default getNConss returns the number of constraints of the transformed problem.

[Opt-Mucca]

Two things:

• This branch was built on your older branch that you changed (I assume). All the display problem stuff should be removed.
• You shouldn't have a default name for the indicator cons. I like the two line change on changing the name if name=="" though and think it's needed

[Joao-Dionisio]

@Opt-Mucca yep, I made an oopsie when branching.

As for your second point, the indicator constraints now have the same behavior as normal constraints - default name = "", and changing it to name = 'c'+str(SCIPgetNConss(self._scip)+1) if this is the case. You are saying that the name should be a required argument? also for the constraints?

[Opt-Mucca]

It is not the same as normal constraints. You have changed the default name to "indicatorCons", which will bypass your change to cX as a default name.

Edit: So my request change would be to have name="" in the function definition as in other functions.

Double edit: I don't know how the tests are passing...... Have I misunderstood something?

[Joao-Dionisio]

@Opt-Mucca I don't think so, I think I changed the default name from indicatorCons, which was there before, to "".

[Opt-Mucca]

I am blind and an idiot.............