state_idempotency

Each state value is treated as idempotent (re-evaluation/re-bootstrapping of state value must yield the same value).

Because of that (for optimization), the evaluated value may be cached. Caching ensures idempotency "de facto" (at least, within current python process), nevertheless, it must be followed "de jure" because proto_code re-starts itself using different python_executable.

If the state has to have different values evaluated for different moments within bootstrap_process, this has to be two different states (different names) and implemented accordingly.