Docs/pretend reading

[adolgert]

This PR is responsible for ordering the documentation correctly and excising repeated documentation. It is not responsible for getting all the words right in each section and is not responsible for having every necessary section.

I cut a lot that looked repeated or unhelpful. I added sections on GSMP, hierarchical samplers, and distributions with memory.

Does this cut too much? Would you recommend a different order? We can do that.

[slwu89]

@adolgert this is great. I think this is the last step before submission to the general repo. I just made a commit which makes a few small edits 662cde4

• add wikipedia ref to variance reduction in CRN section
• in gsmp.jl, whatever method Literate/Documenter uses to parse latex doesn't like \mbox, changed to \text
• in guide.md, slight copy-editing, and added the Haas reference
• in hierarchical.jl, removed an apparently extraneous sentence
• in index.md, a copy-edit, and remove the last line after the last markdown list, which for some reason Literate.jl interpreted as wanting to insert an empty code block in the produced .html file