fix #537 ValueError race condition when running multiprocessing with describe1d

[chanedwin]

References issue #537

Problem :
ValueError is raised when running ProfileReport on large datasets and with multiprocessing on (pool_size >1). This is likely due to the series.fillna(np.nan, inplace=True) in summary.py seems to be performing multiple in-place mutations to the underlying DataFrame object through the passed series reference, resulting in some kind of race condition where two of the processes try to write to the DataFrame at the same time and the ValueError then occurs. This is also why changing the pool_size to 1 fixes the issue, and why the error doesn't always occur - you probably need enough data and threads to hit the race condition.

Solution :
Replace series.fillna(np.nan, inplace=True) with series = series.fillna(np.nan) , negating any side effects from mutation.
Write test case for multiprocessing describe1d to test for multiprocessing functionality

[codecov]

Codecov Report

Merging #549 into develop will decrease coverage by 0.10%.
The diff coverage is 94.73%.

@@             Coverage Diff             @@
##           develop     #549      +/-   ##
===========================================
- Coverage    87.73%   87.62%   -0.11%     
===========================================
  Files          120      121       +1     
  Lines         3115     3152      +37     
===========================================
+ Hits          2733     2762      +29     
- Misses         382      390       +8     

Flag	Coverage Δ
#examples	31.05% <0.00%> (-0.02%)	⬇️
#issue	76.63% <94.73%> (+0.02%)	⬆️
#unit	84.86% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
tests/issues/test_issue537.py	94.44% <94.44%> (ø)
src/pandas_profiling/model/summary.py	89.28% <100.00%> (+0.04%)	⬆️
tests/issues/test_issue437.py	60.00% <0.00%> (-40.00%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fa25e44...8c68ab7. Read the comment docs.

[chanedwin]

Double checked the Travis, it is failing test_issue147.py on the reset_index() call (Error - cannot convert float NaN to integer), which is actually a bug in pandas 1.1.0 listed here - pandas-dev/pandas#35657

On pandas 1.1.0, I'm getting a ValueError exception when calling dataframe.reset_index() under the following conditions:

Input dataframe is empty
Multiindex from multiple columns, at least one of which is a datetime
The exception message is ValueError: cannot convert float NaN to integer.