add tests to check that a Slave Instance will not be allocated when a Software Instance is destroyed

Implement destroy_requested state and transition and mark as soon as possible
this state in case when software instance is being destroyed. In
SalePackingList_requestStoredSoftwareInstanceState support destroy_requested,
which in case of ongoing destruction is no-op.

In such scenario no hosting packing lists shall be created.

When instance has packing list related with setup and cleanup and both are,
confirmed state in case of typical scenario (node says stopped and just after
destroyed) its automatically created instance hosting sale packing lists shall
be delivered.

Do not call checkConnected, in order to have only checkCyclic fired up.

Thanks to this it is possible to manipulate not finished (not delivered yet)
packing lists during destruction process.

vifib_base does not have this portal type, it exists only at vifib_open_trade.

rename variable to be consistent with ComputerPartition.request and follow last changes in person_slap_interation_workflow

refactor code to the paremeters be consistent with ComputerPartition.request. Use shared and shared_xml instead of slave_xml

software_instance was not available anymore.

During finding root software instance not full tree is analysed. In order to
have more complete graph just add some nodes which shall be. They will be
mostly leaf nodes, so they do not impact graph connectivity nor cycleism.

add steps to validate the states of sale packing lists when exists many slave instances and is requested start/stop of all instances

change the parameter of "ERP5Site_getSlaveInstanceListFromComputerPartition" to be possible to pass parameters to catalog search

Root software instance requests shall be disallowed, period.