refactoring bootstrap stats about cinnabar

[RiesBen]

These line looks like a problem to me, for the uncertainty sampling in the bootstrapping.
The random picking of one, is only adding additional noise, not as expected more certainty of the given mean??? why not picking more samples?

cinnabar/cinnabar/stats.py

Line 125 in d96573e

y_true_sample[i] = np.random.normal(loc=y_true[j], scale=stddev_true, size=1)

[IAlibay]

I'm not sure I understand your point fully here @RiesBen - that line should be aiming to add a random noise along the error bar so that your sample accounts for it?