A second instance of a model can be loaded concurrently

[tazlin]

The current set of rules and assumptions in the code are intended to forbid this. Only exactly 1 process should have exactly 1 instance of a model loaded at any given time. However, user reports have revealed that its possible for a model to already be loaded and then be preloaded onto a second process. This can be associated with an immediate unload from the original process or, at least in one case, the second instance doing nothing for at least one job length.

• There may be a race condition with the process map leading to this
• There may be a hole in the preloading check logic