Comment 24 for bug 168695

OK I've committed r12959. to move the remaining asserts to return_ifs. There is a transactional race condition happening still which is infrequent, but is related to the fix put in place. Basically half way through a transaction the button is released and tries to set_sensitive which fails.

But I'm not sure what should be done. Block the call, refuse to set sensitive if there's a transaction taking place. Bit of a catch 22 since most solutions have consequences.