Backport health-monitor changes to 1.3 series drivers

[Cedev]

We're having problems with the 1.4 series driver on icehouse. I think the migration code tries to load lbaas v2 code regardless of whether lbaasv2 is present.

Backport the health monitor changes to 1.3 so they can be used with icehouse.

[Cedev]

Sniff tested with lbaasv1 and lbaasv2 using 2.7.2 and 4.1 boxes.

When lbaasv1 tries to disassociate a health monitor from a 4.1 box, it doesn't. We send

POST /axapi/v3/slb/service-group/4b0652e5-760a-43ac-9ff9-09446c767f29 HTTP/1.1
Host: 10.48.5.198
Content-Length: 94
Accept-Encoding: gzip, deflate
Accept: */*
User-Agent: ACOS-Client-AGENT-1.2.7
Connection: keep-alive
Content-type: application/json
Authorization: A10 491b9e02267d5a3fe8b06b41ec517d

{"service-group": {"health-check-disable": 0, "name": "4b0652e5-760a-43ac-9ff9-09446c767f29"}}

This is generated from the call

        c.client.slb.service_group.update(pool_name, health_monitor="",
                                          health_monitor_disabled=True)

The box responds with

{
    "service-group": {
        "reset-on-server-selection-fail": 0,
        "protocol": "tcp",
        "name": "4b0652e5-760a-43ac-9ff9-09446c767f29",
        "backup-server-event-log": 0,
        "traffic-replication-mirror-sa-da-repl": 0,
        "uuid": "274aad54-a2ad-11e5-9344-fa163e04a247",
        "stats-data-action": "stats-data-enable",
        "sample-rsp-time": 0,
        "traffic-replication-mirror-ip-repl": 0,
        "lb-method": "round-robin",
        "stateless-auto-switch": 0,
        "member-list": [
            {
                "member-priority": 1,
                "uuid": "2a3c54cc-a2ad-11e5-9344-fa163e04a247",
                "name": "_0203f_1_2_3_4_neutron",
                "a10-url": "/axapi/v3/slb/service-group/4b0652e5-760a-43ac-9ff9-09446c767f29/member/_0203f_1_2_3_4_neutron+80",
                "member-state": "enable",
                "port": 80,
                "member-stats-data-disable": 1
            }
        ],
        "priority-affinity": 0,
        "traffic-replication-mirror-da-repl": 0,
        "extended-stats": 0,
        "traffic-replication-mirror": 0,
        "health-check": "fa08674b-f11a-4975-81e7-5094",
        "traffic-replication-mirror-sa-repl": 0
    }
}

The service group is still associated with the health-check.

I suspect that disassociating a health monitor from a service group in the 3.0 api is different than in the 2.1 api and the same error exists in master too.

[Cedev]

The same issue exists in master. #155