Refactor cube testing objects/mocking with DummyCube

[truth-quark]

This closes #69.

This PR covers multiple elements to switch tests to using DummyCube:

• Replaces several Mock objs masquerading as cubes with DummyCube
• Removes old setup cruft
• Cleans some test fixtures
• Adds coords to other test fixtures (fixtures better open for reuse)
• Removes mock.patch("umpost.um2netcdf.fix_latlon_coords") in process() tests
• BONUS: removes mock.patch("umpost.um2netcdf.fix_level_coord") in process() tests

Any comments welcome!

[marc-white]

Looks pretty straightforward to me.

My main comment (which only just occurred to me, and clearly is an existing issue) is that a lot of the test assert statements don't have an explanatory note after them, e.g.:

assert False, "Well this will obviously fail"

It's nice to get an explanation out of pytest as to why a test failed, rather than having to dig into the test code itself to figure out what's going on. Up to you whether this is dealt with as part of this PR, or as a new issue.

[truth-quark]

My main comment (which only just occurred to me, and clearly is an existing issue) is that a lot of the test assert statements don't have an explanatory note after them, e.g.:

assert False, "Well this will obviously fail"

It's nice to get an explanation out of pytest as to why a test failed, rather than having to dig into the test code itself to figure out what's going on. Up to you whether this is dealt with as part of this PR, or as a new issue.

Some of this is related to the process of retro-fitting testing & the way the code operates. Quite often, a function takes inputs, modifies a cube & returns None. Error messages could make it clearer if a value hasn't been set properly.

This relates to another element: being unfamiliar with why steps are done, thus not knowing what meaningful error msg to provide (or in some cases, if the code is doing the right thing).

There's also a future refactoring step where I suspect simplification could result in shrinking the code & tests. At this stage I'm mainly focusing on code coverage to assist with the rework steps.

[truth-quark]

My main comment (which only just occurred to me, and clearly is an existing issue) is that a lot of the test assert statements don't have an explanatory note after them, e.g.:

assert False, "Well this will obviously fail"

It's nice to get an explanation out of pytest as to why a test failed, rather than having to dig into the test code itself to figure out what's going on. Up to you whether this is dealt with as part of this PR, or as a new issue.

I've added this to #27 for feature fixes too.

[blimlim]

Nice! Looks good to me!