Doing an unblock of a Code observer triggering a step over the breakpoint and
immediately (before the step over the breakpoint is completed) requesting the
deletion of the Code observer from that address will hang the inferior.

This is most likely related to bug #4383 which explains that we send too many
stop events to the inferior.

I wrote a testcase
frysk.proc.TestTaskObserverCode.testMultiTaskUpdateUnblockDelete that shows the
problem. It can probably be simplified a little.

The behaviour of the above test is slightly different between x86_64/fc6 (kernel
2.6.22.9-61.fc6) where it hangs frysk itself (!) and x86/f8 (2.6.23.1-31.fc8)
where only the inferior hangs, but frysk-core seems fine (and the tests times out).