Add/Modify the PropertyMap stage on new solution

[captain-yoshi]

I would like to be able to add a property (which cannot be computed in advance) after a solution is found. E.g. compute the transform between a robot link and a attached collision object subframe, which would be needed by a cartesian controller.

Here are my 4 approaches to fix this. You might have other ideas.

Alpha

I could go down the PredicateFilter route and add a new wrapper that add's a callback for changing the PropertyMap on new solution.

Beta

It might be interesting to add an optional callback to the NoOp stage to #534.

Gamma

It would be nice if a callback could be set to any stage.

Delta

One could add the callback directly in the PropertyMap. As I'm writing this, I think this is the cleanest solution. No change in the API and the callback can be triggered only for the chosen solution (post-planning), as opposed to be triggered on every possible solutions while planning.

[rhaschke]

Probably adding a post-processing callback somewhere is the way to go. I don't yet get the idea behind your last proposal: What do you mean by adding a callback in PropertyMap? Storing a "post-processing" callback as a property?

[captain-yoshi]

Yes storing a "post-processing" callback as a property.

[rhaschke]

Then we are on the same page. This is a reasonable approach.