This is technically their bug, not ours, and I think IPHON's fork of their featuralizer fixes it. There's a file of notationally complex segments that get special treatment in featuralization (e.g. /kp/, /ts/), but there's a separate list of notationally complex segments that the featuralizer tries to hit the special-treatment file for. /tɕ/ is in the first list, but not the second, so although there's a handcrafted featuralization for it, that handcrafted featuralization never gets loaded.
(When I say it's fixed in our fork, I mean I'm pretty sure I added all the segments with artisanal featuralizations to that second list.)
But since this is obscure behavior that impacts search, maybe it'd be better to leave it as is -- and (if this is possible; their URLs are a little weird) link to PHOIBLE's own pages for the segments, to make it easy to see what the featuralization is.
This is technically their bug, not ours, and I think IPHON's fork of their featuralizer fixes it. There's a file of notationally complex segments that get special treatment in featuralization (e.g. /kp/, /ts/), but there's a separate list of notationally complex segments that the featuralizer tries to hit the special-treatment file for. /tɕ/ is in the first list, but not the second, so although there's a handcrafted featuralization for it, that handcrafted featuralization never gets loaded.
(When I say it's fixed in our fork, I mean I'm pretty sure I added all the segments with artisanal featuralizations to that second list.)
But since this is obscure behavior that impacts search, maybe it'd be better to leave it as is -- and (if this is possible; their URLs are a little weird) link to PHOIBLE's own pages for the segments, to make it easy to see what the featuralization is.