Add tests

[mtazzari]

We would need to add a few tests, at least:

• to test the package can be imported
• to test the objects can be initialised correctly

Ideally:

• to test that Frank produces the same result on the same dataset (perhaps fixing the random seed we can obtain reproducible results to a numerical precision level?)

These tests would be run automatically after any push command.

[mtazzari]

Some tests placeholders added in #10
Now need to add real code inside them

[mtazzari]

2 tests still missing:

• Test IO
• Test the full pipeline

[jeffjennings]

For the IO test, in tests.py there's

def test_import_data():
    # TODO
    pass

but how useful is this, as it'll just be checking that np.genfromtxt and np.savetxt / np.save are working?

For the full pipeline test, there are already tests for the geometry fitting and the brightness profile fitting; is there utility in a test for the full pipeline, as it would just additionally be testing that figures can be generated (that np.genfromtxt and plt.savefig are working)?

[mtazzari]

Regarding the IO test, now there is a meaningful test here

def test_import_data():
    """Check the UVTable import function works for a .txt"""
    load_uvtable('tutorials/test_datafile.txt')

[mtazzari]

Regarding the full pipeline test, I agree it would be nice to have it -- indeed I added that in Dec to the todo list above.
However to implement such text will require non-trivial amount of work to check files are created, populated correctly, etc. It could easily take a few days of work.
I've opened #48 to track this.