Add "best size" encoding/decoding methods

[SeanTAllen]

When writing a value that is part of a U16, U32 etc, they value might possibly be represented in a smaller value. Add methods/functionality to allow this to happen.

Note, this should be a separate method so any additional processing overhead is not incurred if speed of encoding is valued over compactness of representation.

This might require changes to decoding as well to take a value that was encoded as a U8 but we are expecting as a U16 to get it back as a U16.