terminate request during instance start-up can SEGFAULT the NC (shows up under heavy load)

If a terminate request arrives for an instance that is being started, the NC may receive a SEGFAULT, thus bailing on any activities/transfers going on at that moment. Although this can be reproduced with a single instance that takes a long time to start, this can also happen for small, fast-booting instances that take a long time to start due to disk contention (e.g., when 16 instances are started simultaneously, each from a different EMI).

------------------------------------------------------------
revno: 936
committer: Dmitrii Zagorodnov <dmitrii@z2>
branch nick: 1.6
timestamp: Thu 2009-10-22 01:15:16 -0700
message:
  By adding a new state, CANCELED, fixed the race between startup_ and monitoring_thread, which can lead to SEGFAULT if a terminate request arrives while startup_thread is running.
------------------------------------------------------------