Filter api-gateway cache logging to reduce log output on server disconnect

[curtbushko]

Changes proposed in this PR:

• I noticed this when I was working on a customer issue and it made debugging more difficult
• I was working on an issue that was reproduced by deleting the consul-server-0 pod and the api-gateway logging spit out about 20k log lines in about 5 seconds
• I don't think we need to log these errors out at the api-gateway level as other parts of the controller will log a similar error out but less frequently.
• The customer will still see that the consul server is unreachable in the logs (see below)

Instead of 20k of these:

2023-08-31T14:33:54.861Z ERROR error fetching config entries for kind: http-route {"error": "Get "http://10.244.0.8:8500/v1/config/http-route?dc=dc1&index=10\": dial tcp 10.244.0.8:8500: connect: connection refused"}

They still see:

2023-08-31T14:41:43.031Z ERROR Reconciler error {"controller": "endpoints", "controllerGroup": "", "controllerKind": "Endpoints", "Endpoints": {"name":"consul-ui","namespace":"consul"}, "namespace": "consul", "name": "consul-ui", "reconcileID": "ce66b001-0173-40d1-95ab-a7a5363a6612", "error": "1 error occurred:\n\t* Get "http://:8500/v1/catalog/nodes?dc=dc1&filter=Meta%5B%22synthetic-node%22%5D+%3D%3D+%22true%22+&ns=%2A": dial tcp: lookup : no such host\n\n"}

How I've tested this PR:

• build locally, delete consul-server-0 and look at the connect injector logs.

How I expect reviewers to test this PR:

👀

Checklist:

• Tests added
• CHANGELOG entry added

[t-eckert]

This is great! Thank you

[t-eckert]

I think you will want to backport this into 1.2.x