Burrow reporting client_id and owner empty

Nobody has claimed this yet.

We are currently using
* Burrow: 1.5.0
* Kafka: 2.8.1
* `kafka-version` in `burrow.toml`: 2.8.1

We currently do not get `client_id` and `owner` from the rest api. Both values are empty.

```
curl http://localhost:8000/v3/kafka/CLUSTER/consumer/CONSUMERGROUP/lag | jq '.status.partitions[] | {partition: .partition, status: .status, owner: .owner, client_id: .client_id}'

{
"partition": 0,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 1,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 2,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 3,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 4,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 5,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 6,
"status": "OK",
"owner": "",
"client_id": ""
}
{
"partition": 7,
"status": "OK",
"owner": "",
"client_id": ""
}
```

With another installation with the following specs this seems to be working fine:
* Burrow: 1.3.4
* Kafka: 2.8.1
* `kafka-version` in `burrow.toml`: 2.2.0

Is there anything we need to change to get the `client_id` and `owner` reported correctly?

Reproduce the v3 consumer lag response with the reported Burrow and Kafka versions, using the kafka-version setting from burrow.toml. Trace the endpoint that returns status.partitions and compare it with the working Burrow 1.3.4 setup. Done means client_id and owner are populated correctly for the affected consumer group.

Written by the indexing model from the issue text.