[BUG] Broadast of scalar inputs to the same lengths as other columns in DataFrame constructor

[shwina]

In the constructor of cudf.DataFrame, when one or more input "columns" are provided as scalars, they are broadcast to the same length as other input columns:

df = cudf.DataFrame({'a': [1, 2, 3], 'b': ['x', 'y', 'z'], 'c': 4})
print(df)

Output:

   a  b  c
0  1  x  4
1  2  y  4
2  3  z  4

However, if the scalar appears as the first value in the input, cuDF complains about a length mismatch:

df = cudf.DataFrame({'c': 4, 'a': [1, 2, 3], 'b': ['x', 'y', 'z']})  # ValueError