Testing assert functions give confusing message on failure

[olliesilvester]

assert_reading and assert_configuration give a confusing full diff when an equality fails, making it hard to see the actual issue.

As an example, if we change test_assert_reading in tests/core/test_signal.py so that the dummy_reading has timestamp: 5, this is the test output

AssertionError: assert {'mock_readable-float_array': {'value': array([  1.1231,  -2.3   , 451.15  ,   6.6233], dtype=float32), 'timestamp': 366659.164213077, 'alarm_severity': 0}, 'mock_readable-int_array': {'value': array([1, 2, 4, 7], dtype=int8), 'timestamp': 366659.164199919, 'alarm_severity': 0}, 'mock_readable-int_value': {'value': 188, 'timestamp': 366659.164185914, 'alarm_severity': 0}} == {'mock_readable-int_value': {'alarm_severity': 0, 'timestamp': 5, 'value': 188 ± 1.9e-04}, 'mock_readable-int_array': {'alarm_severity': 0, 'timestamp': <ANY>, 'value': approx([1 ± 1.0e-06, 2 ± 2.0e-06, 4 ± 4.0e-06, 7 ± 7.0e-06])}, 'mock_readable-float_array': {'alarm_severity': 0, 'timestamp': <ANY>, 'value': approx([1.1231 ± 1.1e-06, -2.3 ± 2.3e-06, 451.15 ± 4.5e-04, 6.6233 ± 6.6e-06])}}
  
  Common items:
  {'mock_readable-float_array': {'alarm_severity': 0,
                                 'timestamp': 366659.164213077,
                                 'value': array([  1.1231,  -2.3   , 451.15  ,   6.6233], dtype=float32)},
   'mock_readable-int_array': {'alarm_severity': 0,
                               'timestamp': 366659.164199919,
                               'value': array([1, 2, 4, 7], dtype=int8)}}
  Differing items:
  {'mock_readable-int_value': {'alarm_severity': 0, 'timestamp': 366659.164185914, 'value': 188}} != {'mock_readable-int_value': {'alarm_severity': 0, 'timestamp': 5, 'value': 188 ± 1.9e-04}}
  
  Full diff:
    {
        'mock_readable-float_array': {
            'alarm_severity': 0,
  -         'timestamp': <ANY>,
  -         'value': approx([1.1231 ± 1.1e-06, -2.3 ± 2.3e-06, 451.15 ± 4.5e-04, 6.6233 ± 6.6e-06]),
  +         'timestamp': 366659.164213077,
  +         'value': array([  1.1231,  -2.3   , 451.15  ,   6.6233], dtype=float32),
        },
        'mock_readable-int_array': {
            'alarm_severity': 0,
  -         'timestamp': <ANY>,
  -         'value': approx([1 ± 1.0e-06, 2 ± 2.0e-06, 4 ± 4.0e-06, 7 ± 7.0e-06]),
  +         'timestamp': 366659.164199919,
  +         'value': array([1, 2, 4, 7], dtype=int8),
        },
        'mock_readable-int_value': {
            'alarm_severity': 0,
  -         'timestamp': 5,
  +         'timestamp': 366659.164185914,
  ?                      ++++ +++++++++++
  -         'value': 188 ± 1.9e-04,
  ?                     ----------
  +         'value': 188,
        },
    }

Really, there is only one value causing an issue. It would be easier to write and fix tests if the output here only contained the breaking differences.

[olliesilvester]

Issue rewritten from what I'd originally put since I thought the full diff was actually causing the errors

[coretl]

I think this is contributing to the problem:
pytest-dev/pytest#3638

A possible fix is to allow the user to omit timestamp and alarm severity from the expected values, and if not present then put the actual values into the expected values so the diff is smaller.