uniform_range_coverage_sample() with log_space=True returns incorrect number of samples

[Thomas Gilgenast]

when calling lib5c.util.sampling.uniform_range_coverage_sample() we expect this last condition to be True - that is, we expect to get the number of samples n that we ask for

>>> import numpy as np
>>> from lib5c.util.sampling import uniform_range_coverage_sample
>>> data = np.exp(np.random.normal(size=1000))
>>> n = 100
>>> centers = uniform_range_coverage_sample(data, n, log_space=True)
>>> centers.shape == (n,)
False

it does work as expected if log_space=False

[Thomas Gilgenast]

this test case is actually non-deterministic:

>>> np.bincount(
...     [uniform_range_coverage_sample(np.exp(np.random.normal(size=1000)),
...                                    n, log_space=True).shape[0]
...      for _ in range(1000)])[99:101]
array([337, 663], dtype=int64)