Investigate Integrating Flatbuffers instead of NanoPB

[bperseghetti]

Look at how this will work in ZROS where we just point to memory location, should get much more performance with flatcc.

Look at if shared memory and no need to deserialize is worth the change.