[BUG] loc and iloc indexing with boolean mask does not handle invalid length correctly

[wence-]

Describe the bug

When the indexing array passed to iloc (or loc, which I don't like because it's not label-based in this case) is boolean, the idea is that it should select those rows for which the mask is true [NA values are treated as false, which AFAICT is an unnecessary stipulation because there's no way to construct a pandas boolean array with NA values in it].

If the length of the mask array does not match the length of the frame, we should obtain an IndexError. However, cudf produces an empty frame in this case. EDIT: I might have had an old cudf, so I can't reproduce, but cudf raises RuntimeError rather than IndexError here.

Steps/Code to reproduce bug

import pandas as pd
import cudf
import numpy as np

s = pd.Series([1, 2, 3])
s.iloc[np.asarray([0, 1], dtype="bool")] # => IndexError
c = cudf.from_pandas(s)
c.iloc[np.asarray([0, 1], dtype="bool")] # => RuntimeError

Expected behavior

Should probably match pandas.