fix container addressability issues with cloud-init, precise, when lxc-clone is true

Extensive live tests on MAAS and EC2 with addressable LXC containers revealed a few issues, which need fixing before releasing 1.23-beta1:
1) bad interaction between dnsmasq from lxc (lxcbr0's DNS and DHCP server) - static IPs getting replaced after a 10.0.3.x IP lease assigned via DHCP from dnsmasq expires and gets renewed;
2) unreliable networking and/or slow startup of containers due to race conditions between cloud-init and other networking-related init jobs (especially with precise). All of those issues are exacerbated by enabling lxc-clone (enabled by default), i.e. bad UX.

After several days of testing on both EC2 and MAAS, a patch for all of those issues is about to be proposed.