Make shadow set optional

[dhendryc]

The shadow set is useful if the LMO is expensive to evaluate.

But for the SimpleBoundable LMO, calling the LMO can be faster than going through the shadow set and picking a best point from there.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

❗ No coverage uploaded for pull request base (main@72fbb65). Click here to learn what that means.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #154   +/-   ##
=======================================
  Coverage        ?   86.11%           
=======================================
  Files           ?       17           
  Lines           ?     1520           
  Branches        ?        0           
=======================================
  Hits            ?     1309           
  Misses          ?      211           
  Partials        ?        0           

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

[matbesancon]

this was always returning false then??

[dhendryc]

Rather, it threw an error because the compiler did not know the flag.
In the MOI case, we have a separate strong branching function, which is the same as before.

[matbesancon]

Seems one on the tests should be loosened to 1e-2

Buchheim et. al. mean risk: Test Failed at /Users/runner/work/Boscia.jl/Boscia.jl/test/mean_risk.jl:64
  Expression: dot(a, x) <= b + 0.0001
   Evaluated: 16.076037872611014 <= 16.073528148796374
Stacktrace:
 [1] macro expansion
   @ ~/work/Boscia.jl/Boscia.jl/test/mean_risk.jl:64 [inlined]
 [2] macro expansion
   @ /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.6/Test/src/Test.jl:1151 [inlined]
 [3] top-level scope
   @ ~/work/Boscia.jl/Boscia.jl/test/mean_risk.jl:31
Test Summary:              | Pass  Fail  Total
Buchheim et. al. mean risk |    1     1      2

[dhendryc]

Yes. I think this might be due to SCIP not sticking to its own tolerances.