Split track augmenter into IP and Grade functions

This is a cherry-pick from the p3600 branch which I want to keep in master.

We should split the track augmenter into parts that can be applied to independent of the jet (the grade, in this case) and things that depend on the jet (the signed track IP). The first we can run in derivation jobs so we don't have to store this information in the b-tagging object.

The signed IP information depends on the jet axis, so we should never actually be stored on a track. As the code stands now we're still stashing the information there before running the track writer, but this is a bit sloppy because it relies it relies on overwriting the same track information if we use multiple jet collections.

@mproffit, if you don't see a problem with this I'll merge it.