test case for out-of-order drop

[kazcw]

Demonstrates issue #6

[mehcode]

Thanks for writing up a test for this. Perhaps a keyed ReentrantMutex from parking_lot could be an answer?

[kazcw]

This bug isn't related to threading, so it can't be fixed with a mutex. It's inherent to the current API, which doesn't enforce the requirement that guards that may be for the same address are dropped in the reverse of the order in which they are created. I have a proposed approach to fixing it in #6.