Support for numpy.allclose

[SimonBoothroyd]

It seems pint > 0.10 loses support for numpy.allclose. Running

from pint import Unit
import numpy as np

x = np.array([1, 2, 3]) * Unit("kelvin")
np.allclose(x, x)

yields

TypeError: no implementation found for 'numpy.allclose' on types that implement __array_function__: [<class 'pint.quantity.build_quantity_class.<locals>.Quantity'>]

I'm guessing this was just missed by #905? Happy to submit a PR / test if this is the case.

[keewis]

I don't think np.allclose was ever supported. The only change between 0.9 and 0.10 is that it does not fall back to the magnitude anymore. This:

In [3]: x = np.arange(5) * ureg.m 
   ...: y = x.to(ureg.mm) 
   ...: np.allclose(x, y)                                                                                                                                                                                                       
.../pint/pint/quantity.py:1377: UnitStrippedWarning: The unit of the quantity is stripped.
  warnings.warn("The unit of the quantity is stripped.", UnitStrippedWarning)
Out[3]: False

shows the problems with that approach.

That said, I think it was just missed before.

[hgrecco]

It would be good to use this into account when tackling #947