GitLab

This is because of history/diffmode.

... since we only get one at a time.

sql is such a fuss...

Rather than just crashing ...
This typically happens when there is a discrepancy between
some installed lib version and a required version within
a dependancy.

This should be a one-shot operation.

Since straight out of the db we only get an offset,
and the utc timezone is quite implicit. We want it to be explicit
since pandas can change it interpretation from version to version
(e.g. .23 to .24 in a .loc in the presence of a datetime with offset
but no tz).

(we know it is utc because:

 * postgres is configured (see conftest) to UTC
 * we insert as utc
)

It will help to seamlessly assemble queries out of
* plain string
* query parameters
* pre-built `sqlp` string + parameters elements

`sqlp` carries an sql string fragment plus its needed parameters

... rather than just a name list.

Since supervision uses one, we'd better do this right.
Was lots in the transition from sqlalchemy to plain sql.

We still keep the stripped changesets in db because it is basically
impossible to remove them using the current schema.

The foreign keys from series tables to the changeset table
turn it into a neverending operation.

Fix for this will belong to #54.

This highlight an issue: we don't have the name of the series of the
stripped changesets.

These orphans are created when strip is invoked.
Next, strip will use it.