Update to JuMP v0.23 and switch from Cbc to HiGHS

[odow]

No description provided.

[ccoffrin]

I did notice that JuMP got back into the package deps. The objective of this PR was to remove it, #207

[blegat]

It's only in the test dependencies

[codecov]

Codecov Report

Merging #232 (9dd85ae) into master (ec81a10) will increase coverage by 0.62%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #232      +/-   ##
==========================================
+ Coverage   91.68%   92.31%   +0.62%     
==========================================
  Files          20       20              
  Lines        1961     1991      +30     
==========================================
+ Hits         1798     1838      +40     
+ Misses        163      153      -10     

Impacted Files	Coverage Δ
src/MOI_wrapper/MOI_wrapper.jl	88.83% <100.00%> (+4.09%)	⬆️
src/debug.jl	100.00% <0.00%> (ø)
src/filter.jl	100.00% <0.00%> (ø)
src/bb_user_limits.jl	100.00% <0.00%> (ø)
src/bb_inits_and_defaults.jl	100.00% <0.00%> (ø)
src/table_log.jl	95.31% <0.00%> (+0.03%)	⬆️
src/fpump.jl	89.62% <0.00%> (+0.04%)	⬆️
src/bb_integral_or_branch.jl	97.22% <0.00%> (+0.07%)	⬆️
src/util.jl	73.78% <0.00%> (+0.16%)	⬆️
src/solver.jl	79.76% <0.00%> (+0.24%)	⬆️
... and 4 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ec81a10...9dd85ae. Read the comment docs.

[odow]

It's nice to see that HiGHS passes all the tests.

[ccoffrin]

It looks to me this was the only breaking change that impacted Juniper in the move from MOI v0.10 to 1.0? Is that correct?

[odow]

It's a change in how JuMP manages its backend.

I don't think we ever encouraged people to call .optimizer.model.optimizer, so this is arguably a bug-fix. We should have used direct_model if we cared about the backend. The new unsafe_backend provides a nice way to get the inner model for cases like this.