PICMI: Simulation input meta-data standardisation for Particle-in-Cell(PIC) codes

Publishing raw PIC-simulation results according to the FAIR principle is difficult,
primarily size with a single large simulation's raw data reaching 100s TeraByte up to
1-2 PetaByte.
Fulfilling the FAIR principles therefore requires a different approach, with only
compressed results and simulation setup/initial state made directly available in direct
access public data bases.
This assures findability of simulations if initial conditions are automatically searchable, makes raw data on slow, high capacity archive storage accessible for specific simulations of interest, or allows rerunning the same simulation if additional data is required.

This is currently hindered by the different historically grown input description standards of different PIC-codes.

Simulation setups can neither be parsed, searched or understood without implementing a dedicated parser for each code and with more than 4 PIC-simulation codes, (PIConGPU, WARPX, Smilei, PICLS, ...) in use at HZDR alone this is unfeasible.
To realise the above a common standardized description for PIC-codes is needed. PICMI is being implemented and developed by HZDR and Berkley Labs for this purpose.
Besides fulfilling the above requirements, PICMI will also allow reuse of user interfaces between codes, make the simulations more accessible to users and allow using a common setup for different codes thereby al Bytelowing easy direct comparisons between codes for better reproducibility as well as lay the ground work for automated machine learning on simulations.