Update specification.rst

[ryanmwhitephd]

Replaced taxonomy Rule 10, which appears to violate our intention to decouple the taxon name from the measurement process. It also leaves some situations unclear as to whether to use a parameter or add a token.

The optimal differentiation probably requires some fundamental modeling thought to definitively answer.

Potential differentiating criteria in the meantime: Likelihood of additional parameter values in the future--maybe Parameters for numeric input and influence quantity values only--no Required-parameter set--yes, drafted accordingly

We should examine examples to break, refine or support this rule.

[ryanmwhitephd]

Updated Rule 10 in docs branch.