Reproject fails with error when self.res contains numpy floats

[adehecq]

Describe the bug
When running Raster.reproject on an object for which the items in self.res are numpy.float64 (as opposed to float), it fails at this line with "ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()"
This is because the test dst_res == self.res[0] == self.res[1] behaves differently. With float, it returns a single boolean, for np.float64, it returns a tuple of booleans.

To Reproduce
As self.res cannot be set manually, the best way to reproduce (and the way I came across the bug) is to load with downsampling:

import geoutils as gu
rst = gu.Raster(gu.examples.get_path("everest_landsat_b4"))
print(type(rst.res[0]))
# -> float

rst = gu.Raster(gu.examples.get_path("everest_landsat_b4"), downsample=2)
print(type(rst.res[0]))
# -> numpy.float64

rst.reproject(dst_res=(20, 20))
# -> ValueError

Expected behavior
The function should proceed without error.

System (please complete the following information):
Irrelevant