Refactor infer_variable_types for Dask entities

[thehomebrewnerd]

In infer_variable_types() in entity_utils.py there is a len() call to get the dataframe length. This call causes the entity creation process to be slow for Dask dataframes. Additionally, this function also contains a .compute() call on the sample dataframe, but this computed sample dataframe is never used for Dask as the user must specify the datatypes for Dask entities.

This code could be refactored with these changes:

• Only perform the len() call if the input dataframe is a Pandas dataframe
• Remove the code block that computes the sample_df if the input df is a Dask dataframe as this sample is never used for a Dask entity
• Revert the code for selecting the sample to match the code on master. This code was updated to work with Dask dataframes, but since sample_df is no longer needed for Dask, this can be reverted to its original form.

[thehomebrewnerd]

Closed by #957