Keep implications between restarts

[Opt-Mucca]

This PR changes the data structure holding implication information to a HighsHashTree (similar to vbs). It now transfers implication information between restarts, shrink-problem calls, and sub-mips. This has been a long-standing TODO in the code, but with low priority because it's not going to be too performance changing.

Pros:

• We now save implications from a cut or conflict that were not copied over between restarts. We also save implications from columns that may not have been re-probed on.
• Implications are now valid throughout the entire presolve code and can be used in more complicated reductions down the line.
• Can now add implications from multiple places without corrupting the data structure

Cons:

• There's some overhead to all the new data structures (probably minor, but still).

Testing locally I'm getting a 0.5% slowdown but solving two more instances. I'd say this is mostly noise and I'd wait for @fwesselm to try it on a larger test set.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 73.20%. Comparing base (1945ea8) to head (76e6732).
⚠️ Report is 19 commits behind head on latest.

Additional details and impacted files

@@            Coverage Diff             @@
##           latest    #3209      +/-   ##
==========================================
+ Coverage   73.17%   73.20%   +0.02%     
==========================================
  Files         445      445              
  Lines      107802   107876      +74     
  Branches    17258    17282      +24     
==========================================
+ Hits        78887    78970      +83     
+ Misses      28639    28630       -9     
  Partials      276      276              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fwesselm]

Looks good @Opt-Mucca!

[Opt-Mucca]

This change has come out as performance neutral (0.01% performance improving), but is mostly there to enable future presolve rules to use any implication information. Therefore it should be merged.