loading a dfg is susceptible to changes in yaml config files

[dehann]

Some factors may have user defined configuration files, that although saved in the jld's, change over time and are no longer compatible with the new computational types. Usual options include saving the IIF/RoME/Caesar git SHA and tell the user go there. What else can we do?

cc @mc2922

[dehann]

FYI @GearsAD

Bigger discussion is how is computational fg vs unpredictable parameter configs are managed for storing a factor graph object which may be recomputed later. Sensor physical or electronic calibration/parameters is a clear example.

[dehann]

will also impact #187

[dehann]

will be impacted by #143