Cardinality regarding eml-methods should be corrected in two places.
eml-dataset has an optional child element methods with the cardinality zero-or-
ONE. It should be zero-to-MANY. A dataset may be qualifed by more than one
corpus of methodological effort, and it cannot be assumed that those efforts
form a single logically and temporally continuous sequence of methodStep.
Also, methods itself should be a SEQUENCE, not a CHOICE, of methodStep (itself
repeatable), sampling (optional), or qualityControl (optional). It is an error
to present a choice of optional elements.