[BUG] Missing data with length-one sliced-based __setitem__ and singleton array

[wence-]

Describe the bug

Trying to set a length-one slice of a series with a length-one array (rather than a scalar) loses the value.

Steps/Code to reproduce bug

import cudf
import cupy
s = cudf.Series([1, 2, 3], dtype=cupy.int64)
s.iloc[:1] = cupy.asarray([7], dtype=cupy.int64)
print(s)
# 0    1 <------ expecting this to be 7
# 1    2
# 2    3
# dtype: int64

Expected behavior

Setting a length-one range with a length-one array should work.

Why does this happen?

There's an off-by-one in copying.copy_range:

    if abs(target_end - target_begin) <= 1: # should be `<`.
        return target_column