maint and Common CSM, fix MixtureRelative

[dehann]

Hi @Affie , note I moved some CSM functions into a common_consolidate file. Most of these functions should be agnostic to the solver being used, and likely a few more in CSM.jl that could move over too. Just wanted to move some over to get the story started. Check if you can add these to the take! model system perhaps please. If both solvers enter and exit CSM the same, and have a few functions in common, then that really helps limit the scope of differences between fetch and take models.

ref #855

[codecov-commenter]

Codecov Report

Merging #929 into master will decrease coverage by 1.12%.
The diff coverage is 58.90%.

@@            Coverage Diff             @@
##           master     #929      +/-   ##
==========================================
- Coverage   62.66%   61.53%   -1.13%     
==========================================
  Files          50       52       +2     
  Lines        4952     4971      +19     
==========================================
- Hits         3103     3059      -44     
- Misses       1849     1912      +63     

Impacted Files	Coverage Δ
src/ApproxConv.jl	95.10% <ø> (-0.29%)	⬇️
src/CliqueTypes.jl	88.88% <ø> (+41.83%)	⬆️
src/CompareUtils.jl	16.66% <0.00%> (-33.34%)	⬇️
src/FactorGraph.jl	70.26% <0.00%> (ø)
src/IncrementalInference.jl	80.00% <ø> (ø)
src/JunctionTreeTypes.jl	22.14% <ø> (-0.23%)	⬇️
src/TreeDebugTools.jl	16.73% <0.00%> (+16.73%)	⬆️
src/Deprecated.jl	7.89% <13.88%> (+2.22%)	⬆️
src/SolverAPI.jl	69.51% <16.66%> (-2.65%)	⬇️
src/CSMCommon_Consolidate.jl	45.71% <45.71%> (ø)
... and 14 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 8bede36...fe8908b. Read the comment docs.

[Affie]

I tried to find the common points and so far it looks like the best places might be:

• blockUntilChildrenHaveStatus_StateMachine <==> waitForUp...
• waitChangeOnParentCondition_StateMachine <==> waitForDown...

blockUntilChildrenHaveStatus_StateMachine fits well and the 2 are (can be) roughly equivalent up to there.

[dehann]

Hi Johan, that sounds right. Probably also

• slowIfChildrenNotUpsolved <==> waitForUp