Add Elastx CI (TF OpenStack)

[Miouge1]

What this PR does / why we need it:
Since the OVH voucher expired 10th of May, ELASTX (one of the leading Swedish Public Cloud provider) donated CI resources for Kubespray on OpenStack.

This PR adds a simple ubuntu18-calico CI job (it used to be on OVH CI), CoreOS image is currently not available on ELASTX, but that's not a problem since CoreOS is going EOL end of May anyway.

Big thanks to ELASTX @holmsten and @bl0m1 to make this possible! ❤️

Notes:
Changing the order of filters from kube_network_plugin_multus|bool|default(false) to kube_network_plugin_multus|default(false)|bool, to address the situation where kube_network_plugin_multus is undef as well as if kube_network_plugin_multus is a non-boolean

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Miouge1

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [Miouge1]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[Miouge1]

It would be great to get some feedback on this from @bl0m1 or @holmsten

A couple of potential points:

• I used v1-standard-2 since that's the smallest image with 2 cores (8GB of RAM). Kubespray with 1 core is possible, but requires to reduce the CPU request to get all pods scheduled.
• AZ used sto1, no clue if there is a preference there?
• Neutron Router is created by TF, it's also possible to create the Neutron Router manually and specify router_id if you prefer?

/hold

[huxcrux]

It would be great to get some feedback on this from @bl0m1 or @holmsten

A couple of potential points:

• I used v1-standard-2 since that's the smallest image with 2 cores (8GB of RAM). Kubespray with 1 core is possible, but requires to reduce the CPU request to get all pods scheduled.
• AZ used sto1, no clue if there is a preference there?
• Neutron Router is created by TF, it's also possible to create the Neutron Router manually and specify router_id if you prefer?

/hold

v1-standard-2 is totally fine with us. AZ doesn't really matter all of them are identical :)

Regarding router(s) It would probably be better to manually create one and reuse for all tests.
This since we limit each tenant to one router meaning CI test could fail in case multiple ones are running at the same time.

[floryut]

Should we keep tf-ovh_xxxx ?

[Miouge1]

I would like to get OVH CI back since ELASTX and OVH are very different OpenStack environments, we can clean it up later if it doesn't work out.

[Miouge1]

I ran into problems with dependencies between TF modules (subnet not finished to create when creating the VM etc...), since there is no support for depends_on on modules, I set -parallelism=1 as a quickfix.

[Miouge1]

@bl0m1 and @holmsten it took a while to get things sorted out (namely the errors around kube_network_plugin_multus), but I think we're good now.

Really interesting to see the nuances with each OpenStack Public Cloud

/hold cancel

[floryut]

Ok all good then, good job
/lgtm