Project init CLI

[tailhook]

I propose to add a command:

edgedb project-init

That reads additional config file in our normal project structure:

# dbschema/project.toml
[dev]
instance-name = "myproj"
edgedb-version = "1.0"

Does the following:

1. Installs specified version of edgedb
2. Create an instance named myproj
3. Runs edgedb -Imyproj migrate to initialize the schema on that instance

Then we can have additional features:

1. Use the file in curl https://sh.edgedb.com | sh pipeline: Ask to initialize that instance if it's run from the project directory
2. Add database_dump="./dev_data.dump" that is restored before edgedb migrate. So that user has both prepopulated data and up to date schema.
3. Ask if user wants to use cloud or install edgedb locally

Note: this proposal has a nice side-effect that edgedb-version used is not mentioned at a random place in readme (if at all) and that we can search for such configs on github to find out what version at least part of ours users use.

[elprans]

I like the idea in general. Would project-init create the initial structure in an empty project, or should there be two commands?

[tailhook]

I think single command is okay when the project is run by itself. But I'm not sure that's good in curl | sh pipeline, since that can be run in arbitrary place. We may try to detect if it's within a project probably by checking .git directory or similar.