Reformulate LeafRefValidation in terms of PathExpression

Current implementation of LeafRefValidation relies on parsing the path expression, containing a duplicate ANTLR parser for that purpose.

Leafref paths are already being fully parsed by the YANG parser, exposing the expression in terms of normal steps.

This should be sufficient to eliminate duplicate parsing and re-formulating LeafRefValidation such that it will walk the individual steps and construct its enforcement structure accordingly.