Canonicalize blank nodes within stanza

[jamesaoverton]

It would be useful to relabel blank nodes using some sort of content hashing strategy. The advantage would be that reading the same stanza twice would result in exactly the same rows.

Since we're already collecting stanzas, this could probably be simpler than the general strategies in development:

• https://github.com/iherman/canonical_rdf
• http://json-ld.github.io/normalization/spec/