Considering how fundamentally difficult it is to solve this issue
(potentially requiring a whole new approach to do it safely), I wonder if it
would be acceptable to leave it unresolved for the time being and add it to
the list of known issues (notably in the PCuABI spec).

yes, leaving it unresolved makes sense.

(robust mutex still works within a process, and the libc can define
_POSIX_THREAD_PROCESS_SHARED to 0 and then pshared mutex is not
required for conformance. i dont know how commonly actual code relies
on this feature, but probably not common.)