Fix starting PCI-having HVMs on early system boot and later

1. Make sure VMs are started after dom0 actual memory usage is reported
to qmemman, otherwise dom0 will hold 4GB, even if just a little over 1GB
is needed at that time.

2. Request only vm.memory MB from qmemman, instead of vm.maxmem. While
HVM with PCI devices indeed do not support populate-on-demand, this is
already handled in libvirt XML.

The later may often cause VM startup fail on systems with 8GB of memory,
because maxmem is 4GB there and with dom0 keeping the other 4GB (see
point 1) there is not enough memory to start any sych VM.

Fixes QubesOS/qubes-issues#3462

linux/systemd/qubes-vm@.service

@@ -1,7 +1,7 @@
 [Unit]
 Description=Start Qubes VM %i
 Before=systemd-user-sessions.service
-After=qubesd.service
+After=qubesd.service qubes-meminfo-writer-dom0.service
 
 [Service]
 Type=oneshot

qubes/vm/qubesvm.py

@@ -1244,10 +1244,6 @@ class QubesVM(qubes.vm.mix.net.NetVMMixin, qubes.vm.BaseVM):
                 stubdom_mem = 0
 
             initial_memory = self.memory
-            if self.virt_mode == 'hvm' and self.devices['pci'].persistent():
-                # HVM with PCI devices does not support populate-on-demand on
-                #  Xen
-                initial_memory = self.maxmem
             mem_required = int(initial_memory + stubdom_mem) * 1024 * 1024
 
         qmemman_client = qubes.qmemman.client.QMemmanClient()