Flow cover cuts

[Opt-Mucca]

This PR adds flow cover cuts to HiGHS. All logic related to the cut can be followed from the HighsCutGeneration::tryGenerateFLowCoverCut function. Currently the cuts are only generated based on rows in the LP, i.e., "aggregations" of size 1.
The performance results are extremely finicky, but seem to be promising, and are especially so for some network based energy problems I've been testing on. This shouldn't be merged before any additional computational experiments are done.
@galabovaa I changed the two tests because (1) the instance now solved at the root node and therefore has the optimal solution but has status interrupted (2) the primal heuristics now jumped quicker to the optimal solution, so I had to up the objective limit to still catch the event.

[jajhall]

Is this PR still active @Opt-Mucca ?

[Opt-Mucca]

@jajhall It's still active, but I've been failing to make them anything but performance neutral on-and-off for months now. @fwesselm has been benchmarking them whenever I try something new and I think the change is significant enough.
I'm happy to close the PR if you don't want clutter. I'm hesitant to merge them with the parameter off by default because I don't want to introduce dead code. I also firmly believe this PR is a couple of lines away from improving general MILP performance by a couple of percent and being properly helpful for certain problem classes. I just can't seem to tune them correctly...... (or there's some logic error and I'm weakening them by accident)

[jajhall]

@jajhall It's still active, but I've been failing to make them anything but performance neutral on-and-off for months now. @fwesselm has been benchmarking them whenever I try something new and I think the change is significant enough. I'm happy to close the PR if you don't want clutter. I'm hesitant to merge them with the parameter off by default because I don't want to introduce dead code. I also firmly believe this PR is a couple of lines away from improving general MILP performance by a couple of percent and being properly helpful for certain problem classes. I just can't seem to tune them correctly...... (or there's some logic error and I'm weakening them by accident)

I'm happy to keep it open, but it would be good to merge latest into it so that conflicts don't build up and have CI tests pass

[Opt-Mucca]

I'm closing this PR. I don't see this being merged anytime soon. For some reason (likely bugs or all the lifted cover cuts), HiGHS is not getting the expected performance improvement that other solvers experience. The branch should still very much stay alive, and this should be looked at again after some months.

[jajhall]

Shame, but worth a try