Failure to unblock interactive shell

[hickeng]

Seen while attempting to recreate #5445

Two bundles attached:
before CDE (hoping to get threadstack in useful location)
after CDE

My hypothesis, informed by hung exec test: https://ci.vcna.io/vmware/vic/11546) :
a. we never send unblock so session.started is never modified
b. when the propertyCollector wait times out we do not close the attach streams
c. because the streams are not closed the exec command never exits
d. in the start session case (not exec) then power off operation will disrupt the streams so the client will eventually return.

[hickeng]

@caglar10ur FYI - this is the issue I was asking about.

[matthewavery]

No update on this in about a year. The tests have been reenabled and are currently passing consistently(as far as we can tell). Closing this with the merge of #7969