Yes, this is perfect !
Apparently, the crash happens in glibc while locking a mutex from the mesh repository thread... It is extremely strange, and I'm not seeing it happening here...
Maybe something to do with the fact that the mutex is also locked from a coroutine (the HTTP fetch completion call happening from within a coroutine of the mesh repository thread).
Until I can understand what's going on, I'll revert the "threaded initial joint rigging data caching on mesh skin fetch" optimization which is responsible for this, and will published a fixed 1.32.2.54 later today.