1086 lb refactor

[PhilMiller]

Fixes #1086

[lifflander]

Looks good to me

[PhilMiller]

Failure of clang-8/macosx/mpich was seemingly due to #958

[codecov]

Codecov Report

Merging #1087 into develop will decrease coverage by 0.00%.
The diff coverage is 94.11%.

@@             Coverage Diff             @@
##           develop    #1087      +/-   ##
===========================================
- Coverage    77.66%   77.65%   -0.01%     
===========================================
  Files          672      671       -1     
  Lines        25806    25794      -12     
===========================================
- Hits         20042    20031      -11     
+ Misses        5764     5763       -1     

Impacted Files	Coverage Δ
src/vt/vrt/collection/balance/baselb/baselb.cc	81.76% <ø> (ø)
src/vt/vrt/collection/balance/baselb/baselb.h	100.00% <ø> (ø)
src/vt/vrt/collection/balance/greedylb/greedylb.cc	97.36% <ø> (-0.07%)	⬇️
src/vt/vrt/collection/balance/greedylb/greedylb.h	100.00% <ø> (ø)
.../vt/vrt/collection/balance/hierarchicallb/hierlb.h	100.00% <ø> (ø)
...c/vt/vrt/collection/balance/lb_invoke/lb_manager.h	100.00% <ø> (ø)
src/vt/vrt/collection/balance/rotatelb/rotatelb.h	100.00% <ø> (ø)
src/vt/vrt/collection/balance/stats_msg.h	97.01% <ø> (ø)
...vt/vrt/collection/balance/hierarchicallb/hierlb.cc	88.23% <80.00%> (-0.02%)	⬇️
.../vt/vrt/collection/balance/lb_invoke/lb_manager.cc	82.31% <95.83%> (+1.78%)	⬆️
... and 3 more

[lifflander]

Does this preempt #1059 ?

[JacobDomagala]

I was curious about the usage of lb_instances_ here but i can see that in PR #1059 it makes more sense when you allow LBType::NoLB to be a valid value

[PhilMiller]

This likely leads to #1059 being closed, though I'm not ready to jump on that yet. I just wanted to get these easy and uncontroversial bits merged now, so that they can be built upon for #875, #965, and #1055