Fix issue with datetime_features column mapping

[elenasamuylova]

Description

An error occurs under the following conditions:

• you specify datetime_features in column mapping
• the datatime feature name starts with "prediction"
• you run the Target Drift preset

It fails with an error Cannot calculate drift metric for column 'prediction' with type ColumnType.Datetime. It appears to incorrectly parse the datatime feature as the prediction column.

Example mapping:

column_mapping_1 = ColumnMapping()

column_mapping_1.prediction="prediction"
column_mapping_1.datetime_features="prediction_timestamp_utc"

Example report (fails):

target_drift_report = Report(metrics=[
    TargetDriftPreset(),
])

target_drift_report.run(reference_data=ref, current_data=cur, column_mapping=column_mapping_1)
target_drift_report

However, it works correctly if:

• the datetime feature column does not have "prediction" in its name (for example, it is called timestamp_utc)
• the datatime column is passed as datetime and not datetime_features

Reproducible example https://colab.research.google.com/drive/13hdgU0mSPTiwjq5pAEAgQZApUcM5v3nV

Implementation: Fix the bug + it seems worth checking how datetime_features are treated in other reports.