I believe that problem is caused by static field initalizator.
I’ve spotted that new thread is started only when doInitialize is done (despite thread.Start() is called) – so I suppose that CLR blocks other threads to avoid concurrent access / double field initalization.
To sum up: Newly-created thread is not started by CLR to avoid concurrent access, but main initalization thread waits for child thread to be done what means deadlock.
Edit
@Sebastian proposed (in a comment) the link that may prove my theory: http://blogs.msdn.com/b/pfxteam/archive/2011/05/03/10159682.aspx