Use GitHub to fetch and push docs data

[jenweber]

This is a baby step towards the "next gen" API docs, where our docs are kept in a git repo instead of solely S3. This does not replace the past work done in the next-gen branch, but is meant to buy us some time to finish it for realsies.

To-do:

• [x] diff the v4 generated files with what's in S3, to check integrity
• [ ] Make sure @ symbols are being handled
• [ ] Try running the changes against older versions of Ember/Data
• [ ] Tests 😬
• [x] Choose a deploy strategy
• [ ] Add git pull --tags to instructions
• [ ] Add volta to instructions. Install with skip setup! https://docs.volta.sh/advanced/installers#skipping-volta-setup and add path manually in .zshrc with export PATH=$PATH:~/.volta/bin (order matters, put it at the end)
• [ ] can skip yarn start command - redundant
• [ ] Can skip yarn serve given the changes to the other app
• [ ] run linter and fill in missing name attribute
• [ ] run clean-urls
• [ ] run fix rev index

How to review and test Follow the README. Does it work?

Non-blockers/stretch goals

• This GitHub action is currently pseudocode. It could be merged as such, or removed from this PR.
• Does this almost work for things like canary now?

Architectural changes:

• Instead of pulling docs from unpkg, read them from the ember.js or ember-data GitHub repo. This is much much faster.
• Only process one version at a time. This is much much faster.
• Apply file changes to the ember-api-docs-data repo instead of tmp
• Remove code for syncing to S3. Instead, add the files to https://github.com/ember-learn/ember-api-docs-data
• Remove dependence on AWS keys
• It can be run 100% offline, assuming you have done the prereqs to clone repos and yarn install

Drawbacks of architectural changes

• You cannot process all the docs from the beginning of time in one go.
• You have to clone several repos locally to run it end-to-end locally

Server Notes CNAME api-store.emberjs.com osff.map.fastly.net

Deploy strategies

• Prember! Install prember and write a script that includes all the URLs in the app
• Use a GitHub action to mirror the contents of this repo in S3
• Use a CDN + the git repo as a file server
• Use a CDN + heroko instance as a file server