process pool: add layered approach that helps create dependencies such as input files for processes

`tauler` uses it's own "script" feature which "automagically" creates an input file and passes it to the script. this should probably not be bound to the "low level" process pool, instead it should prepare the target state for the process pool, or sth similar

Start by reading the existing tauler script feature and the low-level process pool implementation to understand where input files are created and passed to scripts. Done should mean a layered approach separates target-state preparation from process-pool execution and supports creating dependencies such as input files.

Written by the indexing model from the issue text.