`ob deploy init` uses too much of `ob thunk`'s checking code

[ryantrinkle]

ob thunk pack needs to be extremely careful not to delete anything the user might want to save, because it deletes stuff that could easily be important. However ob deploy init doesn't delete anything. Therefore, ob deploy init should use the code that figures out the thunk location, but not the code that checks for stashes and such. Checking for uncommitted changes in the working tree is probably still a good idea, so someone doesn't accidentally deploy something they're expecting to match ob run, but which doesn't actually match.