splitAndBalanceMesh fails on hex & prism mesh

[cwsmith]

The attached mesh and model fails in splitAndBalanceMesh during graph construction (IIUC). It seems like there is an issue with the numbering.
mixed.tar.gz

Note, this mesh has two geometric model faces that are matched.

This is same model and mesh as SCOREC/core#173 .

$ mpirun -n 2 splitAndBalanceMesh gs_bumDNS.dmg bumDNS-Smaller.smb 1 2
ENGPAR Git hash 918049a519f0335a6555bf199cd4c9467c8fcd54
model gs_bumDNS.dmg loaded in 0.001543 seconds
mesh bumDNS-Smaller.smb loaded in 0.186153 seconds
number of tet 0 hex 30745 prism 75580 pyramid 0
mesh entity counts: v 69660 e 245640 f 282305 r 106325
splitAndBalanceMesh: /users/cwsmith/lore/develop/EnGPar/interfaces/apfGraph.cpp:361: void agi::apfGraph::connectToPins(int, int, agi::etype): Assertion `temp_counts[i]==pdl[i+1]' failed.

===================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   PID 30183 RUNNING AT pachisi.scorec.rpi.edu
=   EXIT CODE: 134
=   CLEANING UP REMAINING PROCESSES
=   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
===================================================================================
YOUR APPLICATION TERMINATED WITH THE EXIT STRING: Aborted (signal 6)
This typically refers to a problem with your application.
Please see the FAQ page for debugging suggestions

[diamog]

My first guess is that the numbering is screwed up by the matched faces. IIUC for a matched mesh, the "same" mesh face will appear on each of the matched model faces. Will a call to numberOwnedDimension produce the same number for each matched entity? What do calls to getAdjacent return for a matched entity?

I know its not related here, but how does distribution affect this. Will matched entities be on different parts? Does that affect the answers to the above questions?