Refactor Textractor pipeline into multiple pipelines

[davidmezzetti]

Currently, the Textractor pipeline consists of three different components.

• Logic to download/retrieve content
• A file to html converter currently backed by Apache Tika
• A html to markdown converter

This logic should be split up into multiple pipelines.

• FileToHTML pipeline that converts a file to html using a text extraction backend (Tika currently)
• HTMLToMarkdown pipeline that converts html to markdown
• Existing logic to call those pipelines and download data remains in the Textractor pipeline

This will make it easier to add new file to html conversion engines.