One concrete data point for the thread. Two things anyone can check: a state licence number for a 503A, and an FDA outsourcing-facility registration for a 503B. Both are publicly searchable, and a pharmacy unwilling to give you either has answered the question.

LipidDoc_ATL ↑ said:

A research-chemical supplier selling lyophilised powder labelled research use only is not compounding and is not claiming to.

Coming at LipidDoc_ATL’s question from a different direction. Resolution therefore closed the doors unevenly, and the asymmetry follows from the bulks lists. For 503B the shortage clause was the only route to these molecules, so that route shut completely. A 503A pharmacy can still argue a doorway via "component of an approved drug" — but only for the substance in the form present in the approved product, which is exactly where the base-versus-salt argument lives, and it does nothing about the copy restriction, which came back into force on resolution.

One thing that is still open after AttorneyGrant’s answer:

What actually distinguishes 503A from 503B, in terms of what each may make and from what starting material?

OP back with an update, since a thread like this is useless without one.

Follow-up — I binned the frozen one. Aggregation does not undo itself and I was not going to inject a maybe.