Better asserts in `cooper-pair`

[abegong]

There are lots of places in the unit tests where we do this: assert pair.add_dataset_from_pandas_df(pandas_df, 1, filename='etp_participant_data')

This verifies that the code runs without error, but says nothing about the content of the response.

In this case, the response is something like:

{
  "dataset": {
    "id": "RGF0YXNldDoxMzE=",
    "project": {
      "id": "UHJvamVjdDox"
    },
    "createdBy": {
      "id": "VXNlcjo3"
    },
    "filename": "etp_participant_data_orgID-None",
    "s3Key": "ef1de6a6-a3f3-4dc5-9ccd-6c79880f4c0detp_participant_data_orgID-None",
    "organization": null
  }
}

The structure of this response would make a good test case. Most of the specific values (e.g. filename, s3Key, all the ids, etc.) would not.

It would be great to have a function to assert that two nested dictionaries share the same keys. We could use it all over test_pair.py to:

1. Make the unit tests more thorough, but testing WHAT gets returned not just IF something gets returned.
2. Make the unit tests more informative, by including snippets of returned data.

[abegong]

jsonschema could do 1 but not 2.