slapos_subscription_request: cancel existing subscription request if a new trade condition is detected

slapos_upgrader: do not fixup instance tree virtual master if one software instance is in the same virtual master

Ensure people can see projects they used to see before migration

  Whenever the allocated shared (slave) instance is destroyed or stopped, isn't included on the getComputerInformation, so the timestamp can eventually go back in time (so partition skips process).

  This ensures that the partition information gets a newer timestamp (larger them the previous one).

   All timestamps parameters should be int, and be consistent accross
   the parameters.

   The client compare previous/fetched as float, so use string can break
   later on if we introduce a non-numeric value on returned dict.

    Check if bang_timestamp is consistent and updated
    accordingly by the normal usage.

    Whenever an instance is allocated the partition must
    update its bang_timestamp.

    Ensure the implementation is consistent with the precision (1e6) otherwise partition modification date will never
    ever be larger, when compared.

  This ensure that timestamp is updated, and enforces that cache is updated

Instances were not allocated on this project in reality.

It is otherwise difficult for a Project Manager to access it on the panel

Only payable deliveries will generate invoices.

It is needed for libslap.getInformation method compatibility.
It was dropped by nexedi/slapos.core@784e255f

  Conflict resolution in git added a new line while editing the file

   Use recently added API to speed up the rendering.

  After the 25 of the month, the simulation generate extra SPL for discounts,
  pin DateTime to have stable assertions regardless the date the test runs.

   The object will be reindexed regardless with the recent upgrades (due various factors), so this is end up as a duplicated check.

  Only edit if the document isMemberOf

  Include exceptional value (cabo verde) since the title is different
  between the removed object and the ones that was kept.

  Remove all duplication and normalize id/path to prevent further problems.
  Set int_index to 1 on all region categories to facilitate the migration script, to know
      which ones are the "latest". This flagging allow us search for a proper category
  when cleaning up.

  *portal_categories/region/* was added to keep path to prevent categories be removed
  during the upgrade. This should be done (or not) by a migration script later on, since
  it is required to update the related objects.
  We kept a large list, because we want the int_index:1 be set on upgrade,
  so the change only keeps the removed objects, but edit everything else
  remaining.