Handle the possibility that the agent hasn't started yet

[BridgeAR]

Discussed in https://github.com/DataDog/dd-trace-js/discussions/5257

^{Originally posted by adamdavis40208 February 12, 2025} We're noticing that sometimes a new k8s node will get created, a node service placed on the cluster, and that pod spazzing out because the agent isn't available.

DATADOG TRACER DIAGNOSTIC - Agent Error: socket hang up
Received SIGTERM. Shutting down
SIGINT Received, stopping consumer

Sure, we just spin up another pod and try again, but it'd be nice if this could be retried a few times

[BridgeAR]

I think we should support a retry mechanism to gracefully handle cases like that for a startup time.