BentoML

Fyi, this is also a way
```apiVersion: <http://serving.yatai.ai/v1alpha3|serving.yatai.ai/v1alpha3>
kind: BentoDeployment
metadata:
  name: iris-test
spec:
  bento_tag: iris_classifier:djtgsps4lggxplqz
  extra_pod_metadata:
    annotations:
      <http://ad.datadoghq.com/main.checks|ad.datadoghq.com/main.checks>: |
        {
            "openmetrics": {
              "init_config": {},
              "instances": [
              {
                  "openmetrics_endpoint": "<http://%%host%>%:%%port%%/metrics",
                  "namespace": "bentoml",
                  "metrics": [".*"]
              }
              ]
            }
        }
  resources:
    limits:
      cpu: 1000m
    requests:
      cpu: 500m```

<@U0291AW8A7P> could you take a look here?

You've already given the solution, which is to use extra_pod_metadata