Also deploy takeover script only in import instances. This is non-sense to deploy it on export/pbs instances.

It will be used by equeue to prevent to run import scripts if detected.
Reason: if an import script is run AFTER takeover has been done, it may be stopped during its execution while slapgrid is run, causing the new-main instance not to behave correclty.

In case of erp5testnode, we have only one IP. So slapos node instance on clone instance would reach... slapproxy of main slaprunner.

Most important change: revert back develop to false by default in gitclone.

slaprunner: resiliency: allow jinja to read slapparameter_dict and to bubble intsance parameters to main instance.

1/ FTP doesn't work
2/ ftp://ftp.gmplib.org reads FOR FASTER+SAFER DOWNLOADS USE https.

Obvious effect: don't consider processes like mysql as running if it comes from another partition (case of erp5testnode).
Less obvious effect: don't sync/delete supervisord.sock.
Less obvious effect: in erp5testnode, don't uninstall things from main instace when running buildout of instance of clone runner instance.

1/ FTP doesn't work
2/ ftp://ftp.gmplib.org reads FOR FASTER+SAFER DOWNLOADS USE https.

A new parameter 'use-nat', True by default will specify if we will setup a nat interface.
if 'use-tap' is Truei (false by default), a tap interface will be connected with a new interface in kvm.
if 'use-tap' and 'use-nat' are True, then virtual machine will have two interfaces, one with tap and
one with nat.