Address `**` Edge Cases

[Ethan-DeBandi99]

Numpy has variations when using **, power, or sqrt

>>> n = np.array([-0.0, -7, -np.inf])
>>> n**0.5
array([-0., nan, nan])

>>> np.power(n, 0.5)
array([ 0., nan, inf])

>>> np.sqrt(n)
array([-0., nan, nan])

>>> n**(1/3)
array([ 0., nan, inf])

>>> np.power(n, 1/3)
array([ 0., nan, inf])

>>> n**(1/4)
array([ 0., nan, inf])

>>> np.power(n, 1/4)
array([ 0., nan, inf])

[stress-tess]

I think most of the edge cases can be handled by using the chapel sqrt function, so that might be worth looking into a bit