While working on bug 744067, I used this tool as documented to establish new slavealloc entries for those 50 slaves (see attachment to that bug). The import ran and did not report any failure or return non-zero exit status (no listing kept) When bringing the slaves into staging, bld-centos5-32-vmw-036 was missing - examination showed it was skipped on import, as the auto-increment slaveid's go from -035 to -037. I manually added -036 via SQL. As I understand it, many folks are moving to direct sql manipulation of slavealloc, so an appropriate solution may be just to remove the csv import functionality.
WONTFIX in favor of bug 1135818.