monitoring related exporters and tools use API interface network instead of management interface network.

[Gaël THEROND]

Hi team, the monitoring tools within kolla such as prometheus server/exporter do use the api interface network to listen.

The monitoring trafic shouldn't use this interface network but rather its own dedicated network or as a last resort, it should rather use the management network.

Within some secured (public safety/governments) and heavy loaded environments, such mix of API network trafic with the monitoring trafic isn't allowed or is crushed by the API network trafic pressure.

Here is an exemple for the prometheus openstack exporter configuration template:
https://opendev.org/openstack/kolla-ansible/src/tag/18.0.0/ansible/roles/prometheus/templates/prometheus-openstack-exporter.json.j2#L2

Users should be able to select which network they want to use by having the template to use a new monitoring interface variable that can be filled within globals.yml by the user the same way we allow users to attach their network interface / management interface etc.

Thanks!

**Environment**:
* OS (e.g. from /etc/os-release): Irrelevant.
* Kernel (e.g. `uname -a`): Irrelevant.
* Docker version if applicable (e.g. `docker version`): Irrelevant.
* Kolla-Ansible version (e.g. `git head or tag or stable branch` or pip package version if using release): ALL