Add option for Input Table with Athena Linker connection

[StephenBowser]

Is your proposal related to a problem?

While using Splink with AWS Athena it would be great to be able to start with an existing table in Athena rather than a dataframe.

Describe the solution you'd like

When connecting with the linker object it takes an input dataframe and an output Athena connection as shown below. When you run Linker it creates a table in Athena and stores the data in s3. This table is referenced later when running things like linker.training etc..

Example:
import boto3
from splink.backends.athena import AthenaAPI
from splink import Linker, SettingsCreator, splink_datasets

boto3_session = boto3.Session(region_name="eu-west-1")
df = splink_datasets.historical_50k
db_api = AthenaAPI(
boto3_session,
output_bucket=bucket,
output_database=database,
output_filepath=filepath,
)
settings = SettingsCreator( ... )
linker = Linker(df, settings, db_api=db_api)
linker.training. ......

For my use case I already have tables set up in Athena that I want to run matching on. I would like to see an option to pass that table through the Linker() function rather than just having the option for a dataframe. As it is now, I need to read in the data as a dataframe and then it gets sent back to Athena/s3 which is redundant and problematic on larger data sets.

Describe alternatives you've considered

Currently I can read in all of the data that I have in the Athena tables. This is large and redundant.

Additional context