See https://jira.mariadb.org/browse/MDEV-12950 for API changes.

Because this does not seem to be enough:
https://github.com/perl5-dbi/DBD-mysql/pull/133

We should keep using MariaDB 10.1 for ERP5 as long as perl-DBD-mysql
is not fixed properly.

  The version 2.2.0 depends on PyNaCL which is not well configured currently on buildout.

@Tyagov @luke 

This MR adds the following to all Jupyter notebooks:
 - A warning when a user inputs a reference that has already been used in one of his old Jupyter notebooks. 
 - Some extra information when entering magics such as the remaining required magics.
 - Entering the reference MUST now be the last magic entered(otherwise the reference check doesn't work).
 - A small addition to the message at the top of the notebook for the command ```%notebook_set_title```

Also, do note that the edited kernel uses files from [this commit](nexedi/erp5@4a3e5173) from my erp5 repo, so the changes there must be available for the kernel to work. [This merge request](nexedi/erp5!283) addresses those changes.

Sample screenshot:
![Screenshot](/uploads/a34e0b70b991302539c572ad4344ba56/Screenshot.png)


/reviewed-on nexedi/slapos!177

/reviewed-on nexedi/slapos!157

/reviewed-on nexedi/slapos!166

@Tyagov  @kirr 

/reviewed-on nexedi/slapos!171

Buildout commit in discussion:  nexedi/slapos.buildout@43839169

/reviewed-on nexedi/slapos!176

  Follow up the change done on boost-lib: 'component/boost-lib: build with SlapOS gcc.'
  Include --with-boost on configure.

  Use  /folder/script can lead to mistakes which cause error, as
  the instance side cannot find the file. Use directly the section is better to ensure
  you get the file even after changes on common.cfg.

  Clean up location set on common.cfg, and reduce code duplication.

  instance-runner-export.cfg.in extends instance-runner.cfg, so we don't need the secton here.

@Tyagov @alain.takoudjou @Nicolas @kirr 

/reviewed-on nexedi/slapos!158

/reviewed-on !175

Further rewrite should be done but at least there's no awk code anymore.

  Apache MPM Watchdog promise was introduced to search apache on
  deadlock state after a graceful reload and kill them after 600s.

  This change prevents the apache lack of nodes after several processes
  get stalled on deadlock.

  This can be used to define minimal metrics to the frontend to
  prevent mpm nodes starvation after gracefull restarts

  Wait for 60 to reload apache configuration in order to accumulate
  several logrotate runs.

  If the amount of slaves are too high, the number of logs are high,
  so the entries on logrotate are also high. So it is enough to DDoS
  with a huge amount of 'kill -1', so delay is the only way to avoid
  to re-implement logrotate existing features.

  Only reload the apache configuration if the the apache configuration
  or the certificates contains a change, else don't reload it.

  Keep a command on bin folder to force reload of configuration in
  case it is required.

The mpc and mpfr component already in the component/binutils.cfg. Remove the duplicate ones.

Update: the test passed. @rafael 

/reviewed-on nexedi/slapos!163

/reviewed-on nexedi/slapos!173