moving the half fat epi tests into the main half-fat test suite, as it is much faster now and doesnt need to be a discretionary task run

adding logger setup code for epi_processor runs when running fit.py
removing calls to psutils when displaying memory usage

making run dirs be archived if running processor from __main__ (was defualt before, now requires specifying)

updating the default full integration test run date to use the latest production NAME file formats from the Met Office

splitting out logger creation in processor_base to allow processors to be called with a different logger setup (for example epi_fit)