New MARC fields/subfields should come into existence when named in post-processing rules

As noted in comments to ZF-57, we have problems with post-processing rules that fill in the value of one field with that of another, implementing fallback – as for item-level copy-number falling back to holdings-level copy-number. In such cases, the rule never runs because the field it is transforming doesn't exist.

Such fields should be summoned into existence, bowl-of-petunias-like.