The problem has been investigated. And here is the conclusion:
The object in question belongs to the overlap area processed during S13 DC by IN2P3 in patch=366,0. The IN2P3's version of the patch was not included into the merged Stripe82 catalog. The NCSA's version was used instead. User hit a very narrow hole in the merged dataset right on the edge.
Note that this is a well known problem with the current version of the Stripe82 catalog loaded into PDAC/Qserv database sdss_stripe82_00. The problem is well understood, and it will be solved in a refined version of the catalog.
I picked that patch by a pure accident when preparing that tiny monolithic catalog for SUIT in an anticipation of the PDAC deployment. I wouldn't draw any conclusions from this as this monolithic catalog is not meant to be used to Q&A PDAC! For that (later) purpose we're planning to have the full merged catalog on some dedicated hardware....when/if it will be available.