Issues with using numpy arrays of Quantities indexing and inplace operations

[alexfeld]

I often need to store arrays of variables that have different units, e.g. when solving a system of equations. I've noticed some inconsistent behavior with several operations. Note below that Qx is an instance of a quantity, not necessarily with the same units

f = np.empty(3, dtype=object)
x = np.array([1., 2])
f[[0,1]] = x * Q   # f now contains the magnitude of x*Q, not the quantities as I would expect

y = np.array([Q1, Q2, Q3], dtype=object)
y *= Q4  # the magnitude changes correctly but the units do not. The units of the divisor are lost and the units of the original quantities are preserved

These issues can be mitigated using loops and indexing but it would be nice if these operations were supported.

[hgrecco]

It seems that numpy is taking precedence over the pint to execute this code. Maybe there is a work around. I need to take a closer look

[hgrecco]

Revisit after #905