Finalize fixes on WIndows test suite for v0.1.1

[weiji14]

Description of proposed changes

Towards a fully working Windows test suite! Several of our unit tests (~8) have been failing since #434, and we've addressed them in several PRs already. This one will hopefully squash the last of them.

Merge after these are completed (in order):

• Fix logo image discrepancy with conda ghostscript (Let PyGMT work with the conda GMT package on Windows #434)
• Fix Windows not finding 'share' directory (Remove the variable GMT_SHAREDIR from CI settings #437, and related upstream fix)
• Fix Windows on Azure Pipelines not connecting to http://oceania.generic-mapping-tools.org/ (Cache test data on Azure Pipelines #438)
• Fix Windows not finding gmt.dll without GMT_LIBRARY_PATH environment variable (Improve how PyGMT finds the GMT library #440)
• (This PR)
  • Fix sphinx pygmtscraper setting temporary directory on different mount paths (590a523)
  • Fix pandas.Series returning dtype=int64 in Windows repr (efe35ec)
  • Fix issues with c_int/c_long type (9202941)

Fixes #

Reminders

• Run make format and make check to make sure the code follows the style guide.
• Add tests for new features or tests that would have caught the bug that you're fixing.
• Add new public functions/methods/classes to doc/api/index.rst.
• Write detailed docstrings for all functions/methods.
• If adding new functionality, add an example to docstrings or tutorials.

[weiji14]

Last 2 tests to fix! Investigating now, might be related to how numpy on Windows handles integers compared to Linux/MacOS. (see e.g. numpy/numpy#10317?)

================================== FAILURES ===================================
__________________ [doctest] pygmt.clib.conversion._as_array __________________
221 222 Examples
223 --------
224 225 >>> import pandas as pd
226 >>> x_series = pd.Series(data=[1, 2, 3, 4])
227 >>> x_array = _as_array(x_series)
228 >>> type(x_array)
229 <class 'numpy.ndarray'>
230 >>> x_array
Expected:
array([1, 2, 3, 4])
Got:
array([1, 2, 3, 4], dtype=int64)
C:\Miniconda\envs\testing\Lib\site-packages\pygmt\clib\conversion.py:230: DocTestFailure
___________ [doctest] pygmt.clib.conversion.kwargs_to_ctypes_array ____________
263 Returns
264 -------
265 ctypes_value : ctypes array or None
266 267 Examples
268 --------
269 270 >>> import ctypes as ct
271 >>> value = kwargs_to_ctypes_array('bla', {'bla': [10, 10]}, ct.c_int*2)
272 >>> type(value)
Expected:
<class'pygmt.clib.conversion.c_int_Array_2'>Got:
<class'pygmt.clib.conversion.c_long_Array_2'>
C:\Miniconda\envs\testing\Lib\site-packages\pygmt\clib\conversion.py:272: DocTestFailure

[seisman]

Perhaps change the tests a little bit. For example, changing

>>> x_array

to

>>> x_array.dtype
>>> list(x_array)

[weiji14]

c_int is just an alias for c_long according to https://docs.python.org/3/library/ctypes.html#ctypes-tutorial. Changing it to c_long should make this test pass in a cross-platform way.

[weiji14]

Awesome, all green and good to go!