It's not. As I told the author (who contacted me to ask if any further steps were needed), registration after the fact is irrelevant. Publication of nomenclatural acts cannot
be made valid retroactively: the paper
evidence of having been
registered in ZooBank before
it was published. Unlike some other parts of the Code, Art. 8.5.3
is clear and unambiguous:
8.5. Works issued and distributed electronically. To be considered published, a work issued and distributed electronically must
8.5.1. have been issued after 2011,
8.5.2. state the date of publication in the work itself, and
8.5.3. be registered in the Official Register of Zoological Nomenclature (ZooBank) (see Article 78.2.4) and contain evidence in the work itself that such registration has occurred.
Arts. 8.5.1 and 8.5.2 are fulfilled, Art. 8.5.3 is not.
As I also told the author (again), there are two options.
Either the journal has a paper edition after all, in which case the name will be (or perhaps has already been) validly published without any further ado, and registration in ZooBank was not even necessary. (ZooBank is not mentioned in Art. 8.4, only in Recommendation 8A.)
Or the author needs to write a whole new paper which establishes the name by being validly published. Such a paper could be very short; for instance, it wouldn't have to repeat the diagnosis but could cite the old one (or, as I now learn, the Theropod Database!) for that.
Nothing can retroactively make the existing electronic paper validly published. If there is or will be a paper version, the date the paper version is published is the date of publication of the name, because the electronic version simply isn't validly published.
There is no third option.
So I asked again if the journal has a paper version. I haven't received any reaction to that.