refactor of the repo and the docker build

[wreality]

@mikethicke @tzouris @monotasker

OK, so I made a pretty aggressive refactor on the repo to hopefully make deployment/build a little less obfuscated (totally realize these were defaults from invenio).

Major points:

• Invenio modules as git submodules: All the modules are now git submodules so they have a specified version (building the same commit in this repo will result in the same image built)
• Move files for each container (kcworks, kcworks-frontend) into their own directory and structure the directories to better match the final resting place of the files in the container (make it easier to understand what to expect where inside the running container)
• Nginx config refactored to be a little more dry.
• Add traefik to proxy/terminate services in development
• Removed any local / development customizations from the nginx config (and replicated in traefik)
• Simplify the local mounts for docker compose development using the rearranged structure.
• Remove SSL config for kcworks-frontend as the loadbalancer will terminate SSL
• Update readme to reflect (most) of the changes
• Created an "init-container" to handle copying static assets to the static assets volume.
• Renamed the .dev docker compose to just docker-compose.yml for DX.

Remaining:

• Fix an error in the ui container.. I'm guessing I didn't bootstrap something correctly
• Add section to readme for how to trust / persist development certificates
• Make sure local development of invenio modlues is still possible... the directory structure for invenio is not ideal, but it should be certainly doable.
• Add Github actions build process for tagged commits to publish to ECR.
• Do we need an init container to run migrations or anything like that... i.e. what would an "update" require be manually done?

I /think/ in my extremely humble opinion that most of these are improvements and not just preferences, BUT I've not been the ones in the weeds, so please know I won't take offence if something needs tweaks.

[mikethicke]

Thanks Brian, this all makes sense.

My main concern is getting the images pushed to ECR so Dimitris can use them for the production deployment. How close are we on that?

[wreality]

@mikethicke very. The frontend container should be good to go and when I can chat with Ian about the app container error on Monday I'm sure it'll be a quick fix. (If not, the container itself isn't really any different from the original so we should be able to orchestrate and stand everything up one way or another by tomorrow) (bold proclamations, aside)