Improve new pipeline tests

[madwort]

One test of the new pipeline was added in euctr/crawl/tests/test_get_trials_from_db.py - it would be nice to add this type of test for more of this code, to mimic what @NickCEBM was doing in his notebook.
We can generate initial test cases by simply wrapping the helper functions with e.g. input_dataframe.head().to_csv('/path/to/my/file'), but then we would ideally groom the data to exercise more of the code / edge cases / etc.

[madwort]

Nb. the new code loads directly from the database, so be careful with types when loading from CSV - things like bools & dates are converted to python primitives from db but string objects from csv.