Refactor NodeID

[gdbelvin]

NodeID has many different output formats and conversion functions which are scattered throughout the code. It would be great if we could isolate all the formats and conversions to a single API with tests that enforce consistency between the various formats. The idea is that node.ID could be a consistent, easy-to-read package with minimal dependences for all bit and path manipulation purposes.

Proposal:

• new node package to contain node.ID.
  • Should the package be named nodeid?
• Make ID's members private if possible to enforce new API use.
• Use node.ID in place of big.Int
• Shorten NewNodeIDFrom* functions to NewFrom
• Use ID in place of Prefix and Suffix since all three use the format "Path + depth"

Other suggestions?

[Martin2112]

I think this was completed?

[gdbelvin]

This has a lot more work to do to simplify and unify the number of input and outputs to *Node.
However, at this point there's not pressing features that require this simplification. It's more of an enhancement / cleanup than a feature blocker.