This adds tests for the new /items POST handler, both with mocked
data and a live connection to a Solr statistics core. Tests that
only work when Solr is available are marked with XFAIL so that they
don't turn the whole test suite red.
In each test I try to assert as many parameters as we can know for
each response so that we cover all expectations. For example, when
we test a valid limit parameter we should test whether the response
not only has the same limit parameter, but that the number of items
has actually been limited and the number of pages has been adjusted
accordingly.
See: https://docs.pytest.org/en/stable/skipping.html
DSpace 6+ uses a UUID for item identifiers instead of an integer so
we need to adapt our tests accordingly. The Python UUID object must
be cast to a string to use it elsewhere in the code.
Falcon's response content is raw bytes, while its text is a string.
Let's use the latter so we can use json.loads() in Python 3.5, 3.6,
and 3.7 with the same code.
See: https://falcon.readthedocs.io/en/stable/api/testing.html
