`to_compact()` fails with numpy quantities

[sieben-gea]

to_compact() is a no-op on NumpyQuantitys and results in the warning:

UndefinedBehavior: to_compact applied to non numerical types has an undefined behavior.
qty.to_compact()

The problem appears, because for numpy quantities, the qty.magnitude is of type ndarray, which is non-numeric, an therefore qto.to_compact() returns the quantity without change:

pint/pint/facets/plain/qto.py

Lines 95 to 103 in 40936b9

	if not isinstance(quantity.magnitude, numbers.Number) and not hasattr(
	quantity.magnitude, "nominal_value"
	):
	warnings.warn(
	"to_compact applied to non numerical types has an undefined behavior.",
	UndefinedBehavior,
	stacklevel=2,
	)
	return quantity

The method should be modified to accept ndarray by using either the maximum or minimum order of magnitude in the array to select the unit prefix.

Minimal example to reproduce the problem:

from pint import UnitRegistry
ureg = UnitRegistry(force_ndarray=True)
qty = ureg[Any]('1e+3 g')
qty.to_compact() # 1000.0 gram