Now that all states have a dropdown, we can simplify the code.
This also brings the "Repasser en instruction" action for all states
(including when no attestation or motivation is present).
- Fix the missing link to see attestations (see #4400)
- Fix the "Repasser en instruction" wording being confusing for
accepted state
- Add lot of tests
- rename `dossiers_id_with_notifications` to `followed_dossiers_with_notifications`,
- rename `notifications_per_procedure` to `procedures_with_notifications`,
- return an ActiveRecord::Relation instead of the result of the query, so that the call place can compose it,
- `merge` with the wanted Dossier scope in the call places, don’t bother passing it as a parameter,
- use the “state” (now “scope”) parameter as a scope method that can be just applied on `Dossier`.
It was slightly broken in two ways:
- it doesn’t make other instructeurs unfollow the dossier
- if the dossier is unarchived, instructeurs don’t refollow automatically.
* Get rid of the “reopen” event, merge it with “publish” (it’s the same code)
* Remove the “availability” states; “available with brouillon” makes no sense since the brouillons path are always uuids
* Instead of checking if publish can happen, just try it and handle the errors
It is actually the same thing as Gestionnaire.procedures. It already included the procedures with paths as well as the archived procedures, and in production, there were no Gestionnaire for who procedures was returning a different result than visible_procedures (expect for two baddata brouillon procedures with a nil path).
In addition, Procedure.path is now nonnull, which means the Procedure.avec_lien scope is pointless.
Finally, the current spec showed that the only procedure not visible to the gestion was the one he was not assigned to.
`login_as administrateur, scope: :administrateur` doesn’t seem to actually do anything, but testing for the _absence_ of content can’t break.
- Make the test fail if login fails,
- Fix the login,
- Write the opposite test,
- Update the searched content to the actual text used in the partial
The `conversations_count` we previously used counts replies, but also
all conversations that were tagged or re-tagged during the month – with
made counting the actual work spent on user support brittle.
Counting the replies is a better estimation of what we get.
Unfortunately this also removes the filtering-by-tag feature. To
mitigate this, the reports are now scoped to a specific mailbox. This
allows to create extra mailboxes for conversations that should't be
counted in the stats.
