If you're running 'delete backup' to delete only a subset of your RMAN backuppieces (based upon date/tag etc) then 'delete filespace' from the TSM server might be a little heavy-handed (!).
Never seen this before (although if you have got a lot of historical backups to delete, it can normally take a while from my experience), but I'd be curious as to where the 'wait' was here. Am I right in thinking there are lots of sessions opening and closing between your TDP client and TSM server when running the delete? Is the wait time in between sessions are opening and closing, or whilst a session is open on your TSM server? Does a 'list backup;' give you an idea of how many backup objects you're trying to delete?