Update to_dataframe API Docs to focus on schema use

[damccorm]

The API documentation for to_dataframe (https://beam.apache.org/releases/pydoc/2.32.0/apache_beam.dataframe.convert.html#apache_beam.dataframe.convert.to_dataframe) is very sparse. It also focuses on specifying a proxy, rather than relying a schema-aware PCollection as an input. This function is often people's entrypoint into the API, so we should make it very clear how to use it. Let's expand the documentation, and focus on using schemas rather than specifying a proxy.

We should also link to the documentation for to_dataframe in https://beam.apache.org/documentation/dsls/dataframes/overview/#embedding-dataframes-in-a-pipeline

Imported from Jira BEAM-13002. Original Jira may contain additional context.
Reported by: bhulette.

[TheNeuralBit]

See https://stackoverflow.com/questions/72996664 for an example of confusion this causes. proxy argument is being used rather than declaring a schema. We might want to also improve error messages here (e.g. raise an error if the input element_type is not DataFrame).