Comment 2 for bug 518512

The problem is due to the following condition:

client_do - job X
timeout, don't process result
client_do - job Y
get result for job handle X
since client_do reuses data structures (only 1 active job at a time), context for job X is lost
cannot find job X, hit assert()

I think the best fix here is to remove the assert and simply ignore the result packets if it is no longer found in the active client task list.