[r33 jan 10] While reinstalling apisupport, as usual: the deadlock
generator :-) I'm not sure that what apisupport is doing in its reload
thread is completely fine, but one apparent problem in the request
processor is that DataObject.dispose (presumably holding whatever
datasystems-specific locks etc.) winds up causing DataNode to
fireNodeDestroyed, which can enter the node mutex. This looks to me
like an incorrect lock ordering: creating a DataObject (which resulted
in the call to dispose()) ought not try to get the node mutex.

Suggest that DataNode react to DataObject property changes asynch.