Make tests deterministic

[hugobuddel]

Some tests can randomly fail, e.g.

https://github.com/AstarVienna/ScopeSim/actions/runs/7443459990/job/20248391071

================================== FAILURES ===================================
________ TestApplyTo.test_noise_reduces_with_square_root_of_ndit[9-1] _________

self = <scopesim.tests.tests_effects.test_BasicReadoutNoise.TestApplyTo object at 0x0000027593CEF8B0>
ndit = 9, noise_std = 1

    @pytest.mark.parametrize("noise_std", [1, 9, 100])
    @pytest.mark.parametrize("ndit", [1, 9, 100])
    def test_noise_reduces_with_square_root_of_ndit(self, ndit, noise_std):
        dtcr = _basic_detector(width=256)
        ron = PoorMansHxRGReadoutNoise(noise_std=noise_std, n_channels=64, ndit=ndit)
        dtcr = ron.apply_to(dtcr)
        noise_real = np.std(dtcr._hdu.data)

>       assert noise_real == approx(noise_std * ndit**0.5, rel=0.1)
E       assert 3.3188548593454184 == 3.0 � 3.0e-01
E         comparison failed
E         Obtained: 3.3188548593454184
E         Expected: 3.0 � 3.0e-01

scopesim\tests\tests_effects\test_BasicReadoutNoise.py:36: AssertionError

It would be better to make these deterministic.

PoorMansHxRGReadoutNoise can be made deterministic by passing the random_seed parameter, and np.std probably has something similar.

[teutoburg]

Yes, I've encountered cases like this before. Usually pass if run again. But indeed, deterministic should be the goal! Perhaps we can fix this in the course of #298