Health service instances watcher

[jakubdyszkiewicz]

Solves #9
Adds ability to watch healthy service instances.

I extracted requiredValue function to class otherwise I'd have to duplicate it.

The response from /v1/catalog/service/ is different than /v1/health/service/

The /v1/catalog/service/ response:

[
{
    "ID": "4fbf9ea0-c9a8-a0c5-2c84-6509c0da8639",
    "Node": "localhost",
    "Address": "127.0.0.1",
    "Datacenter": "dc1",
...
    "ServiceID": "sample_service_ip_port",
    "ServiceName": "sample_service",
    "ServiceTags": [
      "tag1"
    ],
    "ServiceAddress": "127.0.0.1",
...
    "ServicePort": 8080,
...
  }
]

The /v1/health/service/ response:

[
{
    "Node": {
      ...
    },
    "Service": {
      "ID": "sample_service_ip_port",
      "Service": "sample_service",
      "Tags": [
        "tag1"
      ],
      "Address": "127.0.0.1",
      "Meta": null,
      "Port": 8080,
      ...
    },
    "Checks": [
      ...
    ]
  }
]

(I ... non important part of the responses)

Because of the different responses, I had to create another JsonDecoder.

The test is not perfect. The new PollingConditions can caught registration of healthy service before unhealthy one, but otherwise I'd have to wait arbitrary time which I did not want to do.