Add preprocessing pipeline

[tanaysoni]

Resolves #381

[flozi00]

Looks cool, left some questions and suggestions @tholor

[prairitm]

Hey, great change in the pipeline, as large documents increase reader inference time.
How can we implement the new preprocessing pipeline in api with file upload?

[tholor]

@prairitm we expect to finish this PR today. We will then test the new PreProcessor a bit in practice to see if we need to iterate further on the design. The integration into the file upload API will then probably follow in the next 2-3 weeks, but you could of course already give it a try yourself.