Avoid FutureWarning in ensemble.py

[michaelmckinsey1]

The current code results in the below warning:

FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df[col].replace({numerical_fill_value: None}, inplace=True)

The change in this PR follows the recommendation in the warning to be compliant.

[ilumsden]

This looks good and, in general, is what you should always use when doing operations.

TBH, the inplace argument to pandas functions is completely useless and should be avoided at all costs. This is because, internally, all operations act on copies of the DataFrame or Series. So, even when you use inplace=True, pandas is internally creating a copy of the DataFrame or Series you are operating on.