wip, consolidate factors and manifolds

[dehann]

I'd like to consolidate ManifoldPrior with Prior, but this will need:

• converters to serialize manifolds (I was thinking of just doing it by name , ie shorthand string, and hard coding the converters, one for each manifold). e.g. "SpecialEuclidean(3)" gets persisted, and user code converts it back to whatever complicated type.
• helper constructors so that legacy use of say Prior(Normal()) automatically inserts M=Euclidean(1),

EDIT @Affie , feel free to take over this branch in the morning hours there. I will be working on this again tomorrow.

[codecov]

Codecov Report

Merging #1291 (7943fa3) into master (9b0c5fc) will decrease coverage by 40.61%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master   #1291       +/-   ##
==========================================
- Coverage   40.63%   0.02%   -40.62%     
==========================================
  Files          57      57               
  Lines        4794    4821       +27     
==========================================
- Hits         1948       1     -1947     
- Misses       2846    4820     +1974     

Impacted Files	Coverage Δ
src/ApproxConv.jl	0.00% <0.00%> (-66.55%)	⬇️
src/FactorGraph.jl	0.00% <0.00%> (-64.64%)	⬇️
src/Factors/DefaultPrior.jl	0.00% <0.00%> (-47.62%)	⬇️
src/Factors/GenericFunctions.jl	0.00% <ø> (ø)
src/IncrementalInference.jl	9.09% <ø> (-68.69%)	⬇️
src/entities/GraphConstraintTypes.jl	0.00% <ø> (ø)
src/FactorGraphTypes.jl	0.00% <0.00%> (-100.00%)	⬇️
src/GraphProductOperations.jl	0.00% <0.00%> (-95.24%)	⬇️
src/CalcFactor.jl	0.00% <0.00%> (-79.17%)	⬇️
src/TreeMessageAccessors.jl	0.00% <0.00%> (-76.00%)	⬇️
... and 32 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 9b0c5fc...7943fa3. Read the comment docs.

[Affie]

The manifolds prior part of this is currently in conflict with local changes I have as a wip to allow parametric and non-parametric to work with the same factors.
Can we hold off on consolidating Prior and ManifoldPrior for now?
I think we can serialize in the same way as with variables with a getManifold function if it's easier, but it is currently more convenient to just store the manifold in the factor.

[dehann]

okay, sure holding off on consolidation of Prior. I'm going to clear vecP stuff in RoME in the meantime then.

[Affie]

Hi, I've uploaded my local changes as is so long: #1292

[dehann]

Replaced by 0f7edd4 in #1300