Fix tests of process to work with the new FCIS restructure

[blimlim]

The branch 145/fcis-architecture implements the big restructure of the process and cubewrite functions, as described in #145. The main changes are:

• IO steps are handled in process(), while all cube modification steps are moved to a new process_cubes() function.
• The new process_cubes() function absorbs any cube modification steps previously part of cubewrite().
  Roughly process() becomes the "imperative shell", performing all the reading and writing, while process_cubes() forms the "functional core", performing cube modifications over several modular steps.

The new structure caused the existing end to end tests of process() to break. This PR follows on from work in the branch 145/fcis-architecture-broken-test, in getting the tests to work. The changes are:

• Change the target of the tests to the "functional core" process_cubes()
• Remove mocks of IO steps
• Add one mock of fix_time_coord, which uses cube methods including cube.coord_dims(), cube.transpose and iris.util.new_axis which would be harder and fragile to implement in the DummyCube.
• The existing tests checked that the correct cubes were either dropped or kept by process by checking that the output cubes were the same objects as the input cubes:

  for pc in processed:
          assert pc in cubes

  This doesn't work with the new process_cubes(), as fix_time_coord() and fix_pressure_levels return new cubes rather than modifying in place. These assertions have been swapped to check that the names of the output cubes match what's expected, e.g:

   cube_names = {cube.name() for cube in cubes}
   processed_names = {cube.name() for cube, _, _ in processed}
   assert processed_names == cube_names

The target of this PR is the 145/fcis-architecture branch instead of main. There are several steps before that branch can be merged into main, and hence to try and make the PRs smaller and easier to review, I'll do separate smaller PRs into 145/fcis-architecture, and do a PR into main when all the steps are complete.

Any feedback or suggestions are welcome!

[marc-white]

This should be good, pending a clean-up of possibly redundant test fixtures

[marc-white]

Good to go!

[atteggiani]

LGTM

[atteggiani]

Thank you @blimlim!
Merged.