Python: Example of JIT-compiled Fortran execute from Python

[mlange05]

This PR provides a first example of how the Just-in-Time (JIT) compilation infrastructure from loki.build can be used to on-the-fly compile the CLOUDSC kernel and invoke it form Python with Numpy arrays.

This change requires two main steps:

• Extension to the utilities in the current cloudscf2py sub-package to account for the transposed Fortran array layout.
• A slightly hacky/hand-holding implementation of the JIT compilation and linkage against the common lib.

The latter step is by no means clean, but highlights the challenges for JIT-compiling self-contained IFS-style kernels/parametrisations.

Please note that the linkage to the common sub-library also requires a dynamic library build!

A few other small updates and clean-ups:

• Updated Loki API usage for regenerating the pure-Python kernel via Loki transformation
• Re-organise Fortran sources into a sub-directory and move Python validation into separate sub-package
• Clean up file-read and validation utilities and add simple docstrings
• Add transpose functionality for input reads and validation steps