Decomplect unification in explanations

[alex-dixon]

We should disambiguate between a constraint that merely "gets" or captures a value vs. one that is an equality constraint.

Then an explanation can be less ambiguous:
where <entityid, attribute, value> is <a literal value or an already bound ?variable>
bind to <first referenced ?variable / unbound ?variable>

Without the distinction we use something like "where" for the capture, which doesn't fit.

Second references to bound variables should use "where" as we are then talking about equality.

[alex-dixon]

This analysis is useful if we add plain English explanations back. This was before we imported our entire syntax :)