[BUG] Calling cudf.from_pandas on empty index isn't preserving the dtype for string column

[galipremsagar]

Describe the bug
When an empty index with object dtype is constructed and passed to cudf.from_pandas, the dtype is being reset to float64.

Steps/Code to reproduce bug

In [1]: import pandas as pd

In [2]: s = pd.Series([], dtype='str')

In [3]: import cudf

In [4]: gs = cudf.from_pandas(s)

In [5]: s
Out[5]: Series([], dtype: object)

In [6]: gs
Out[6]: Series([], dtype: object)

In [7]: i = pd.Index([], dtype='str')

In [8]: i
Out[8]: Index([], dtype='object')

In [9]: cudf.from_pandas(i)
Out[9]: Float64Index([], dtype='float64')

Expected behavior

In [9]: cudf.from_pandas(i)
Out[9]: StringIndex([], dtype='object')

Environment overview (please complete the following information)

• Environment location: [Bare-metal]
• Method of cuDF install: [from source]