Changed __repr__ in BinnedSpikeTrain to support quantities<0.12.4

[kohlerca]

The unit test for method __repr__ in conversion.BinnedSpikeTrain fails if quantities is 0.12.3 or below.

The old method was using Quantity scalars inside an f-string. This required the __format__ method implemented in quantities==0.12.4 to generate the correct string representation of BinnedSpikeTrain with magnitudes and units.

This PR changes the string generation by casting the Quantity scalars to string, which uses the __str__ method in Quantity and displays the magnitudes and units as expected. This does not rely on any formatting method and thereby it is compatible with older versions of quantities.

This addresses issue #409.

[coveralls]

Coverage remained the same at 88.82% when pulling 3165aeb on INM-6:fix/binned_spike_train_repr into 165276f on NeuralEnsemble:master.

[ojoenlanuca]

I can approve these changes.
Little side-note: it is maybe IDE dependent, but when i run test_conversion.test_repr in Pycharm to recreate the bug #409, the self.assertEqual() function might expect as first parameter the desired value and secondly the actual value. Because in the respective debug-message, the order is switched to
Expected: string generated by repr()
Actual: hardcoded reference string

[kohlerca]

This is a PyCharm issue, as it is known that its diff tool provides these expected/actual reports in an order different than the one used in several of our tests (https://youtrack.jetbrains.com/issue/PY-26471), i.e., assuming that the expected value is the first. There is a recommendation by Guido van Rossum against forcing this, and instead, refer to the values in a generic form, such as first/second (https://mail.python.org/pipermail/python-dev/2010-December/106954.html). This is actually what the assert function documentation uses.

If running the tests using the command line, the standard != and the points with the differences are shown without any semantics attached:

E       AssertionError: 'Binn[19 chars]t=1.0, t_stop=1.01, bin_size=0.001; shape=(1, [19 chars]rix)' != 'Binn[19 chars]t=1.0 s, t_stop=1.01 s, bin_size=0.001 s; shap[25 chars]rix)'
E       - BinnedSpikeTrain(t_start=1.0, t_stop=1.01, bin_size=0.001; shape=(1, 10), format=csr_matrix)
E       + BinnedSpikeTrain(t_start=1.0 s, t_stop=1.01 s, bin_size=0.001 s; shape=(1, 10), format=csr_matrix)
E       ?                             ++             ++                ++

[ojoenlanuca]

Ok, so then i can approve the pull-request.

[ojoenlanuca]

These changes prevent test_converstion.test_repr()-failure for quantities version <= 0.12.3 and should be adopted.