Update Documentation

[brandenchan]

Let's update our documentation so that the website better reflects the latest developments in Haystack.

The main changes are:

• Describing Haystack as Search instead of QA
• Replacing Finder with Pipelines
• Summarizer
• Translator
• Crawler
• Evaluation Nodes
• Pipeline Yaml config example
• Adding import statements to code examples

Fixes #793

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[tholor]

Great work and many good additions + fixes! Left a few comments. Ready to merge once those are addressed.

For the next documentation update, I see the following topics:

• Introduce a FAQ section (probably as a subpage of "Overview")
• Dedicated section about optimization of speed (this question comes up a lot and would be great to just help users with a single link. Besides document length, it should contain the levers model size and GPU / CPU)
• Reformat code examples so that they don't contain |when copying/pasting them
• Think about a concept for introducing more "code snippets / recipes" (either within API reference or Usage or a central place where you can browse + search 🤔 )
• Usage section for KGQA
• Usage of REST API

For the FAQ part a few questions that came up a lot in the last months:

Q: Why do I get duplicate answers in the results? A: You added your documents twice to the document store. Make sure to call doc_store.delete_all_documents() before running your indexing script a second time
Q: How can I speed up predictions? A: Splitting docs, using GPU, smaller model (we should have a better documentation page just for this topic)
3. Q: Which model to use for German / spanish / ...? A: (maybe we should update our multilingual table in the docs? )
4. Q: How to add / access metadata?

[brandenchan]

Also we should fix the example for the decision nodes in the pipelines usage page

[brandenchan]

The next batch of changes to be addressed by #992