Split out metadata creation from data import in the local files handlers

[npatki]

Problem Description

The local file handlers (for CSV and Excel) currently return both the metadata and data in one go.

data, metadata = handler.read(folder_name='project/data')

While this approach is ok, it is not consistent with other parts of our product and may not be future-proof.

Expected behavior

1. Update the handler.read function (for both CSV and Excel) to just read and return the data. Do not create or return metadata in this function.
2. Create a new function called create_metadata (for both CSV and Excel). This function should take as input a dictionary of DataFrames. It should output a MultiTableMetadata object.

Below is the updated user journey:

from sdv.io.local import CSVHandler

handler = CSVHandler(sep='\t', encoding='UTF') 
data = handler.read(folder_name='project/data')
metadata = handler.create_metadata(data)

Additional context

In the future, we would like to allow flags in create_metadata in order to control whether inference should be done. But this is out of scope for the current issue.

# EXAMPLE only: out of scope
metadata = handler.create_metadata(
  infer_sdtypes=True,
  infer_primary_keys=False,
  infer_foreign_keys=True)