clib.conversion._to_numpy: Add tests for pandas.Series with pandas string dtype

[seisman]

Description of proposed changes

Add tests for pandas.Series with string dtype. Six cases are tested:

1. dtype=None
2. dtype=np.str_
3. dtype="U10"
4. dtype="string[python]"
5. dtype="string[pyarrow]"
6. dtype="string[pyarrow_numpy]"

Neither can be converted to np.str_ directly. Cases 4-6 can be fixed by 01ba317, and cases 1-3 can be fixed by dac7e8e.

In [1]: importpandasaspdIn [2]: importnumpyasnpIn [3]: x=pd.Series(["abc", "defg", "12345"], dtype=None)
In [4]: x.dtypeOut[4]: dtype('O')
In [5]: np.ascontiguousarray(x)
Out[5]: array(['abc', 'defg', '12345'], dtype=object)
In [6]: x=pd.Series(["abc", "defg", "12345"], dtype=np.str_)
In [7]: x.dtypeOut[7]: dtype('O')
In [8]: np.ascontiguousarray(x)
Out[8]: array(['abc', 'defg', '12345'], dtype=object)
In [9]: x=pd.Series(["abc", "defg", "12345"], dtype="U10")
In [10]: x.dtypeOut[10]: dtype('O')
In [11]: x=pd.Series(["abc", "defg", "12345"], dtype="string[python]")
In [12]: x.dtypeOut[12]: string[python]
In [13]: str(x.dtype)
Out[13]: 'string'In [14]: np.ascontiguousarray(x)
Out[14]: array(['abc', 'defg', '12345'], dtype=object)
In [15]: x=pd.Series(["abc", "defg", "12345"], dtype="string[pyarrow]")
In [16]: x.dtypeOut[16]: string[pyarrow]
In [17]: str(x.dtype)
Out[17]: 'string'In [18]: np.ascontiguousarray(x)
Out[18]: array(['abc', 'defg', '12345'], dtype=object)
In [19]: x=pd.Series(["abc", "defg", "12345"], dtype="string[pyarrow_numpy]")
In [20]: x.dtypeOut[20]: string[pyarrow_numpy]
In [21]: str(x.dtype)
Out[21]: 'string'In [22]: np.ascontiguousarray(x)
Out[22]: array(['abc', 'defg', '12345'], dtype=object)

[seisman]

Changes in this PR were initially added in #684 to support string arrays with np.object_ dtype. It's no longer necessary after dac7e8e because the array has been processed by the _to_numpy function when calling vectors_to_arrays at line 1471.

[seisman]

Specifying dtype is not necesary here, but I feel it's good to expicitly tell that here we're expecting a np.str_ array.

[seisman]

This is necessary to support pd.Series string like:

x = pd.Series(["abc", "defg", "12345"], dtype=None)
x = pd.Series(["abc", "defg", "12345"], dtype=np.str_)
x = pd.Series(["abc", "defg", "12345"], dtype="U10")

[seisman]

Again, this line was added in PR #684 and is no longer needed after dac7e8e