Ignore UTF8 encoding differences on comparison

This does a little simplification of the canonicalization and adds the ability to ignore character encoding differences.

See: #638 (closed)