I’m looking to follow best practices with Databricks Auto Loader. When ingesting different file formats, is it considered good practice to always read data as strings first, or is it better to use format-specific readers (e.g., JSON, CSV, binary) from the start?