Well, Im not privy to how gpumax works internally, so I may be missing something, but clearly they already have a robust form of failover: the private pools.
If something goes wrong processing leased shares, just direct the miners back to their private pools. Thats not worse than what it is now, in fact, its the exact same thing - except if it there is no problem, you can process leases even if pirate sleeps .

Im sure there is a good reason why this isnt done, I just wonder what it is.