Removed unreusable reusable code for better readability

[Helveg]

This code was written by me with a lot of initial enthusiasm for pytest during a hackathon experimenting with best patterns for pytest, and it completely missed its mark for providing an easy reusable simulation fixture. I've removed it in favor of straightforward, less complex, much more readable test code.

[heplesser]

nest.set

Suggested change

	nest.SetKernelStatus({"tics_per_ms": 2**14, "resolution": resolution})
	nest.set(tics_per_ms=2**14, resolution=resolution)

[heplesser]

Suggested change

V_m = nest.GetStatus(neuron, "V_m")[0]

[heplesser]

Suggested change

	assert math.fabs(V_m - expected_V_m) < tolerance
	assert neuron.V_m == pytest.approx(expected_V_m)

[heplesser]

Suggested change

	spike_times = nest.GetStatus(spike_recorder, "events")[0]["times"]
	spike_times = spike_recorder.get("events", "times")

[heplesser]

Suggested change

	events = nest.GetStatus(vm, "events")[0]
	times = np.array(events["times"]).reshape(-1, 1)
	V_m = np.array(events["V_m"]).reshape(-1, 1)
	results = np.hstack((times, V_m))
	actual, expected = testutil.get_comparable_timesamples(
	results,
	np.array(
	[
	[0.1, -70],
	[0.2, -70],
	[0.3, -70],
	[0.4, -70],
	[0.5, -70],
	[2.8, -70],
	[2.9, -70],
	[3.0, -70],
	[3.1, -70],
	[3.2, -70],
	[3.3, -70],
	[3.4, -70],
	[3.5, -70],
	[4.8, -70],
	[4.9, -70],
	[5.0, -70],
	[5.1, -69.9974],
	[5.2, -69.9899],
	[5.3, -69.9781],
	[5.4, -69.9624],
	[5.5, -69.9434],
	[5.6, -69.9213],
	[5.7, -69.8967],
	[5.8, -69.8699],
	[5.9, -69.8411],
	[6.0, -69.8108],
	]
	),
	)
	assert actual == expected
	res = vm.get("events", ["times", "V_m"])
	vm_expected = testutil.get_samples_at_nearest_times(res["times"],
	np.array(
	[
	[0.1, -70],
	[0.2, -70],
	[0.3, -70],
	[0.4, -70],
	[0.5, -70],
	[2.8, -70],
	[2.9, -70],
	[3.0, -70],
	[3.1, -70],
	[3.2, -70],
	[3.3, -70],
	[3.4, -70],
	[3.5, -70],
	[4.8, -70],
	[4.9, -70],
	[5.0, -70],
	[5.1, -69.9974],
	[5.2, -69.9899],
	[5.3, -69.9781],
	[5.4, -69.9624],
	[5.5, -69.9434],
	[5.6, -69.9213],
	[5.7, -69.8967],
	[5.8, -69.8699],
	[5.9, -69.8411],
	[6.0, -69.8108],
	]
	),
	)
	assert actual == pytest.approx(expected)

[Helveg]

@heplesser It seems like the tests do rely on get_comparable_timesamples to subselect both the measured and expected samples. One way we could completely remove this util function altogether is by running the old tests and by exfiltrating the exact solutions into separate arrays that we can then simply assert measured == pytest.approx(expected)? We'd probably be duplicating the solutions a couple of times but perhaps that is preferred? I think using this util function leaves a lot of room for errors in the times of samples being masked because they're filtered out in both the measured and expected array.