Description of problem:
When netork exception occurs (such as bug https://bugzilla.redhat.com/show_bug.cgi?id=1163073) pending resoured are not cleared, eventually cause to insufficiant resources in host while trying to add VM.

Version-Release number of selected component (if applicable):
rhevm-3.5.0-0.19.beta.el6ev.noarch
libvirt-0.10.2-46.el6_6.1.x86_64
sanlock-2.8-1.el6.x86_64
qemu-kvm-rhev-0.12.1.2-2.448.el6.x86_64
vdsm-4.16.7.3-1.el6ev.x86_64

How reproducible:
inconsistently, depends on bug 1163073.

Steps to Reproduce:
1. Using vt9 release, add few VMs and run them.
2. In case of VM get stuck in launching state or failed to launch, try to remove VM and add another one.
3. Observe host "Max free Memory for scheduling new VM" or pending resources in engine databse.

Actual results:
resources are not cleared in case of VM start failure due to network exception.

Expected results:
resources should be cleared in case of VM failed to start.

Additional info:
Enging and host logs attached.