The ContentFunctionsDescriptor will fail to parse the field from a scored phrase when calling fieldsAndTerms

For example:

Given a function like content:scoredPhrase(FIELD, -1.5, termOffsetMap, 'foo', 'bar') the call to fieldsAndTerms should return {FIELD}, {foo, bar}, but it will return an empty field set.

Given a function like content:scoredPhrase((FIELD_A || FIELD_B), -1.5, termOffsetMap, 'foo', 'bar') the call to fieldsAndTerms should return {FIELD_A, FIELD_B}, {foo, bar}, but it will return an empty field set.