This is particularly needed for old OS with python2: very old versions of
pip are trying to install latest version of eggs no matter their python2
compatibility whereas pip 20 is making sure to install only python2
compatible eggs.

See merge request nexedi/slapos.core!433

   Revert inclusion Base_getNodeTitleAndDate and use checked_permission to only get title if the user is allowed to see the document.

   Author is optional so keep empty isn't a problem.

See merge request nexedi/slapos.core!434

Until now, the broadcast address was used for the tun interface instead
of the first address of the range.

See merge request !436

See merge request nexedi/slapos.core!435

  We do want to ignore payments that occur after the at_date.

  We want to know if at a certain date, there still unpayed documents older them that date, if the document is payed after, (and grouped) we unconsider it.

   This was missing to remove since we dropped all mass upgrade code and relies exclusively on Upgrade Decisions.

   Upgrade decisions are also tickets (messages are issued), this makes things consistent with the other tickets (Regularisation Requests and Support Requests)

   This allows administrator see issued messages.

   node_title is another person/organisation title, which was raising if an administrator use himself as sender on the event.

See merge request nexedi/slapos.core!431

If computer partition requested state is `started`, slapos node boot will start
partition services on reboot, event if the node is not able to connect to master or
has no internet.

See merge request nexedi/slapos.core!428

See merge request nexedi/slapos.core!430

   ... Unify scripts into Base_getEventList

  In order to use getMovementHistoryList too

   It is used to get the events that follow up to a ticket that aggregates (or sub object aggregates) to an document (usually Compute Node or Instance Tree).

   Reuse Base_getOpenRelatedTicketList to query tickets consistently and extends original goal to also show Upgrade Decisions.

   This tab shows all Tickets related to the computer and to all instance trees that has one instance allocated on it (even if isn't the one that is failing).

   Use default_or_child_aggregate_uid is enough (and simpler).
   Use specific states to be consistent with RSS queries.

   Reduce the number of scripts and refactor the query script to rely on getMovementHistory.

   Before only the latest event was notified, to reduce the list and only to ping the user to access the site. Now all events are considered on the list (not the latest ones) and include some general clean up on base RSS feeds.

  Now the goal is save the configuration and later assert with saved configuration on the actual site.

  This may give us other uses for this tooling, like, save before and after an upgrade (to know the actual delta) and save after an upgrade to know later the local changes done over time.

  This without loose the possibility of audit (since you can upload a file after fetch the configuration from the unit tests)

   We shouldn't keep this in git, but it should be created or handled after an upgrade by the developer.

See merge request !423

See merge request !429

   Events are creates via script (and API) now, instead jio_post.

Add `README.md` instructions explaining how to adapt the migration
tests after upgrading the proxy database version.