Infra: Research a "no options"/"starter" mode.

[ryan-roemer]

OPTION 1: dogs

• [ ] Add https://github.com/FormidableLabs/aws-lambda-dogs to this LEARNING.md as a place to review and try out.
• [ ] Include instructions for creating a temporary branch, naming it, and deploying a full infrastructure. And teardown instructions at the end.
• [ ] Review aws-lamba-dogs for any changes that can help learning.

OPTION 2: Starter mode in this project

Needs some research, but basically something like:

• [ ] Run yarn starter and have all the TF config + files, JS files, and SLS config that are part of an OPTION(NAME) only.
• [ ] Update the SERVICE_NAME to not conflict with existing deployment.
• [ ] Search on all options tokens like OPTION, layer, vpc, canary, custom_role and verify we got everything.
• [ ] Add LEARNING.md notes on everything including how to do a custom SERVICE_NAME, how to provision, then how to tear down at the end.

Some potential things:

• [ ] (Maybe) create a long-running branch for the simple version
  • [ ] Create documentation (and possibly scripts) for maintainers to keep up to date with master
• [ ] (Maybe) create an AWS env for simple version for just deployment (but not experimentation). (We would probably still want folks to rename SERVICE_NAME to a custom name for experimentation).

Some other minor complexities

• [ ] Remove postinstall (or make it accomodate layers stuff disappearing for ease)

Related, but on master

Other things I noticed for master:

• [ ] Maybe hone down comments in TF config files (like listing all options for terraform-aws-serverless) module anyways to start.
• [ ] Combine versions.tf into main.tf for simplicity in (could be a separate ticket).
• [ ] Have both custom_role and custom_roles in OPTION(). Harmonize these to just one. (could be a separate ticket).
• [ ] Look to hone down serverless.yml more if possible (could be a separate ticket).