Initialize the database automatically when starting mquery

[MichalJura]

Your checklist for this pull request

• I've read the contributing guideline.
• I've tested my changes by building and running mquery, and testing changed functionality (if applicable)
• I've added automated tests for my change (if applicable, optional)
• I've updated documentation to reflect my change (if applicable)

What is the current behaviour?

Currently we needed to initilize database by hand: python3 -m mquery.db.

What is the new behaviour?

Now the database initialization is automatic and it uses alembic upgrade head instad of mquery.db so it also applies migrations. Also if the database already exists it will stamp it and create alembic_version for future migrations to be applied via upgrade head

Test plan

For new instance: Create new instance and it will initilize database automatically.
For existing database: Remove (unnecessary for dev containers) web container and set it up again.
Then you can check if there is alembic_version in postgresql.

Closing issues

fixes #402

[msm-cert]

This looks more complicated than I thought it would. I'll take a look at it and check if it can be simplified.

[MichalJura]

This looks more complicated than I thought it would. I'll take a look at it and check if it can be simplified.

The reason behind it is that we don't have alembic_version table.
So for new instances there is no problem as alembic upgrade head will just apply migrations and add alembic_version to the database.
But if there is already a database alemibc head upgrade will result in error in Postrgres.
We could simply check if the database is initilized and then pass, but the new migrations like #412 would have to be applied manually.
On the other hand we migth want to apply them manually(but I don't think we are adding some breaking changes to the database schema?).
But in that case we could remove the stamping part.

[msm-cert]

The reason behind it is that we don't have alembic_version table.

We should, otherwise how would previous migrations be applied?

If previous documentation was giving incorrect instructions that bypass alembic it's unfortunate, but we don't have to accomodate such weird cases (database was not a part of any released version so it's not a critical problem).

Sorry if that was not clear and you had to solve this problem unnecessarily.

[msm-cert]

I still believe the previous version was severely overcomplicated. Pushed my suggestions/fixes.

One important note:

First thing is, chdir is extremely "dirty" thing to do, and in 99.9% of programs should never be called (it changes process' global state and may cause unpredictable problems). But even more importantly, this place is not guaranteed to exist. It only exists in the current docker environment, but the code should work outside of docker too (many deployments work this way). In this case, code is in a package, and we should just package alembic.ini and migrations.