With bd92f809 (slap/hateoas: use new name for portal type
("Compute Node" instead of "Computer"), 2023-11-30) the API changed.

See merge request nexedi/slapos.core!580

   Use a gadget_erp5_page_json_editor rather them gadget_erp5_page_slap_json_form
   Update tests to follow up changes for the new json-editor

See merge request nexedi/slapos.core!585

See merge request nexedi/slapos.core!584

Prevent netifaces from introducing netmasks in the form of 'ffff::/16'.
Once this has been applied everywhere, all other fixes for this can be
dropped.

Currently computer netmask looks like 'ffff::/16' in slapos.xml.
With this commit, this netmask will be cleaned up to just 'ffff::'.

Computer definition file (slapformat-configuration.cfg) may now have:

```
[partition_0]
...
capability_list =
  fast-network
  more_disk_space
  yet another capability

...
```

Capabilities are line-separated and may otherwise contain whitespace.
Whitespace at the beginning and end of each line is stripped.

See merge request nexedi/slapos.core!583

Previous code matched way too many files as belonging to the current
partition and wrongly removed them as obsolete: e.g. when processing
slappart1, all the configuration files belonging to slappart10 would
systematically be deleted, as if they belonged to slappart1 and were
now out-of-date. Same for all partitions starting with 'slappart1',
such as slappart11, slappart12, etc.

This resulted in all services in slappart1X being stopped and started
repeatedly and very often: stopped during processing of slappart1 and
started during processing of slappart1X, with potentially several
minutes of downtime, and a high percentage of downtime.

This is a fixup of db98a521.

This method is important we want to test it

See merge request nexedi/slapos.core!582

  Fixup minor js implementation error

See merge request nexedi/slapos.core!577

  Since we are changing erp5_site_global_id, update cache is mantatory be called often

See merge request nexedi/slapos.core!576

This could cause slapos request to hang forever while fetching the
json schema of an SR, and in particular it could cause buildout to
hang forever in instance processing when requesting a sub-instance.

See merge request !575

See nexedi/slapos.package!187

   This constraint has no impact and you change your environment (ie.: use another memcache server) fix consistency will just break the system.

   This constaint also contains hardcoded values (which is already the setupped) so it dont make much sense keep this constraint around doing nothing.