Make shape of scalar slices NDArray objects to follow NumPy conventions

[FrancescAlted]

This exposes the issue:

import numpy as np
import blosc2

shape = (5, 5)
slice_ = (slice(4, 6), slice(4, 6))

npa = np.arange(int(np.prod(shape)), dtype=np.int32).reshape(shape)
b2a = blosc2.asarray(npa)

# Get a slice with Numpy
npbuffer = npa[slice_]
print(f"numpy {npbuffer.shape=}")

# Get a slice with Blosc2
b2buffer = b2a[slice_]
print(f"blosc2 {b2buffer.shape=}")

The output is:

numpy npbuffer.shape=(1, 1)
blosc2 b2buffer.shape=()

Blosc2 should follow NumPy convention here.