Hipo fixes

[filikat]

Various fixes to HiPO:

• Fix some overflows when computing statistics of analyse phase.
• Free memory used for normal equations, if augmented system is preferred.
• Fix bug in supernode amalgamation.
• Printing of BLAS library moved to HiGHS header, so it is printed when using HiPO without logging.
• Recommend to install Metis branch 521-ts due to better oredring quality on many problems. Update workflows and documentation accordingly.
• Add option hipo_metis_no2hop to control option no2hop of Metis and add warning if the fill-in is large.

[codecov]

Codecov Report

❌ Patch coverage is 75.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 81.06%. Comparing base (978a958) to head (3c956a8).
⚠️ Report is 19 commits behind head on latest.

Files with missing lines	Patch %	Lines
highs/lp_data/HighsOptions.h	75.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           latest    #2624      +/-   ##
==========================================
- Coverage   81.06%   81.06%   -0.01%     
==========================================
  Files         347      347              
  Lines       85228    85232       +4     
==========================================
+ Hits        69092    69095       +3     
- Misses      16136    16137       +1     

☔ View full report in Codecov by Sentry.
📢 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.

[galabovaa]

Great! 🚀

[filikat]

@odow One of the Julia tests if failing. Do you have a C version of this test that I can look at?

[odow]

I think it's a good thing. You now solve the problem at the root node, where previously it required branching. I'll update the Julia test.

[filikat]

Oh, I see.
I thought that the MIP solver still used IPX by default though @jajhall.
Or did I understand wrong?

[odow]

Then it might be because of unrelated changes? Either way, I'll relax the test: jump-dev/HiGHS.jl#306

[jajhall]

Oh, I see. I thought that the MIP solver still used IPX by default though @jajhall. Or did I understand wrong?

Was using IPX by default in two main cases where IPM is (might be) used in the MIP solver: analytic centre (LP solve without basis), but now using IPX by default in the other: when switching to IPM after simplex reaches iteration limit in HighsLpRelaxation::run. This use my have tweaked the run of @odow.

See #2626