Convert assumptions to list

[IgnaceBleukx]

Allows a user to post a set instead of list of assumption variables during the solve call.
More of a user-experience improvement (personally also encountered this multiple times already) and fixes #689

Could be part of a larger discussion on what to do with sets in general, do we want to support them as if they were lists?
E.g., should get_variables(some_set_of_constraints) also work?

[ThomSerg]

Looks good. But was wondering why we don't check if its an iterable instead? Something similar as is done in _flatten(). Not that this would make any noticeable difference. Just a thought, pull request can be merged as is.

[IgnaceBleukx]

Hmm good point, from a user-perspective it might actually be nice if we just took care of any collection.
I'll adapt it so it calls flatlist(assumptions) instead.

[tias]

I'm not convinced... we should have stricter typing, not weaker?

e.g. this should be a List[BoolVarImpl_] or smth...

if a user has a set, they should just do list({a,b})? Or it can be any Iterable, but 'flatten' means also nested lists?

[IgnaceBleukx]

Hmm yeah explicitly typing this might indeed be the better option, we should check for it as well then and raise a proper error

[hbierlee]

I'd be in favor of an Iterable over literals (without flattening).

On one hand, this is weaker typing than having List[..] which is good: it only requires what is needed (least knowledge principle, I believe). In this case, we only require something we can iterate over to get each next assumption, and whether the assumptions come from a list, set, or some expensive asynchronous generator, doesn't matter to us. In this case, the user may have a good reason to keep assumptions in a set, and it's unnecessary/inconvenient/maybe costly to convert this set to a list.

On the other hand, this typing is stronger b/c it does not accept iterators over non-literals: e.g. the iterable should be "flat" (only generate literals, not suddenly generate lists!), because it's not clear what this means. An assumption is a literal, not a list.

[hbierlee]

Also note: in this PR, please also remove ocus_explanations from skipped examples in test_examples and remove TODO comment in the example code in this PR too.

[IgnaceBleukx]

related to #775

[tias]

right, so should be type hinted as Iterable and not Sequence because we don't care about the ordering...

[tias]

rebased, please review @IgnaceBleukx