The frontend software URL is a virtual URL that does not have a
corresponding schema, so fetching the schema is just a waste of time.

Generally speaking, fetching the schema impact performance a lot, so it
would be nice to cache them. In practice most of the requests made
during instantiation are frontend request, so this is a first easy step
to fix the performance impact of fetching the schema.

See merge request nexedi/slapos.core!445

   Force properly the creation of the invoices and assert if links are correctly displayed.

    The value isn't fetch from a getAttachment anymore

Apply suggestion to master/bt5/slapos_accounting/SkinTemplateItem/portal_skins/slapos_accounting/AccountingTransactionModule_getUnpaidInvoiceList.py

   It is required to show them all before continue

   There isnt a need to fetch on every line, since the information comes from options

  On invoices tab, only display invoices Unpaid (missing payments).

  This makes easy the user know what is pending to pay and on the 'List all' tab, he can
  search old invoices (and view them all by date)

See merge request nexedi/slapos.core!444

   Ensure it wont find any unexpected published notification message.

   This fails if the website property is set on Live Tests or whenever you run on project specific context.

See merge request nexedi/slapos.core!438

   Delay a bit before edit for now.

   This contains external min code, that isnt so complaint with JSL tests

     It is not required to a major test on full scenario to test load parameters from text area, neither 2 tests since it cover both cases as before in a much simpler way.

   Replace tv4 by cfworker json schema and ref-parser

    It was replaced

   As exposed by Romain, `parseFromString` does not throw any error if the XML parsing fails. Instead, you have to manually check if there is a `parsererror` element inside the returned document.

   See: https://developer.mozilla.org/en-US/docs/Web/API/DOMParser/parseFromString#error_handling