Add a CardamonRun table to the db

[ohuu]

Explicitly storing individual runs will probably help in features going forward and will probably help to simplify some SQL queries.

Proposal

CREATE TABLE IF NOT EXISTS Run (
    id INT NOT NULL AUTOINCREMENT PRIMARY KEY,
    start BIGINT NOT NULL,
    stop BIGINT NOT NULL
);

and then add FOREIGN KEY to run_id's in other tables.

[ohuu]

We may need to update bits of the code to use this data. E.g. in building datasets and the summary stuff

[ohuu]

actually I'm not sure this makes sense. Cardamon runs are of observations which include multiple scenarios. The only thing that's important is that a scenario was run at a certain time (and possibly in the future, in response to some event - git commit, PR etc). Once a run happens Cardamon doesn't care what other scenarios were run in the same run and it shouldn't help in any queries to know that.

I'm closing this issue until I find a compelling reason to model "runs" in the db.

[ohuu]

We actually need this for the UI, so I'm reopening

[anewmandev]

@ohuu moving back to triage until you make your bloody mind up