We use a single a single tf state per Google project. Instead of slicing up the state into multiple substates, we use modules as much as possible, and use "-tar

We use a single a single tf state per Google project. Instead of slicing up the state into multiple substates, we use modules as much as possible, and use "-target=module.module_name" to test our branches.