add more calibration tests

[mathause]

Currently we only test one single ensemble member for one scenario in

mesmer/tests/integration/test_calibrate_mesmer.py

Line 12 in 5a45665

def test_calibrate_mesmer(test_data_root_dir, tmpdir, update_expected_files):

We should also add tests for one scenario two ens members and two scenarios. The potential disadvantage is that we lock in the current averaging (or maybe it's good, then we have to be explicit about changes).