Fix: stop reentrancy loop when reassigning tech and group

[Rom1-B]

Checklist before requesting a review

Please delete options that are not relevant.

• I have performed a self-review of my code.
• I have added tests (when available) that prove my fix is effective or that my feature works.
• I have updated the CHANGELOG with a short functional description of the fix or new feature.
• This change requires a documentation update.

Description

• It fixes #45821
• Reassigning a ticket's technician and group at the same time could trigger an infinite loop and a Gateway Timeout.
  The old assign removal is now deferred to the moment the new actor is actually added, instead of running synchronously during the ticket update, which broke the re-entrancy.

Screenshots (if appropriate):

[stonebuzz]

pending_remove_assign_users (set here) and keep_new_assign_users (line 288) are stored in $_SESSION, but the handoff they implement only ever spans a single request: pre_item_update() sets the flag, and a nested pre_item_add hook fired later in the same Ticket::update() call consumes it. If that expected follow-up add never happens (a rule blocks the group/tech add, an exception aborts the request, GLPI dedupes the add before pre_item_add fires), the flag is never cleared and stays in $_SESSION across requests. The next unrelated actor addition on that same ticket, possibly much later in the same session, would then unexpectedly trigger removeAssignUsers()