Add support for MSK clusters with kraft metadata nodes

[errm]

Since kafka version 3.7.0 MSK has supported using kraft metadata nodes instead of zookeeper. https://aws.amazon.com/about-aws/whats-new/2024/05/amazon-msk-kraft-mode-apache-kafka-clusters/

When running prometheus-msk-discovery against a cluster in kraft node there is a panic like this:

2025/01/30 12:03:36 http: panic serving 10.3.57.169:45616: runtime error: invalid memory address or nil pointer dereference
goroutine 18 [running]:
net/http.(*conn).serve.func1()
	/usr/local/go/src/net/http/server.go:1868 +0xb9
panic({0x7675e0?, 0xa8f9c0?})
	/usr/local/go/src/runtime/panic.go:920 +0x270
main.getBrokers({0x86e250?, 0xc00013a140}, {0xc00002bf20, 0x5b})
	/src/main.go:131 +0x24d
main.buildClusterDetails({0x86e250?, 0xc00013a140?}, {0x0, 0xc000268730, 0xc000232b40, 0xc000232a80, 0xc000232a70, 0xc000226930, 0xc000226948, 0xc000232b20, ...})
	/src/main.go:140 +0x56
main.GetStaticConfigs({0x86e250, 0xc00013a140}, {0xc00025ba18, 0x1, 0xc0000ef988?})
	/src/main.go:199 +0x26b
main.httpSD.func1({0x86feb0, 0xc00015a000}, 0xc0000efb18?)
	/src/main.go:248 +0xbb
net/http.HandlerFunc.ServeHTTP(0x440480?, {0x86feb0?, 0xc00015a000?}, 0x6457fa?)
	/usr/local/go/src/net/http/server.go:2136 +0x29
net/http.(*ServeMux).ServeHTTP(0xace040?, {0x86feb0, 0xc00015a000}, 0xc000154000)
	/usr/local/go/src/net/http/server.go:2514 +0x142
net/http.serverHandler.ServeHTTP({0xc000150090?}, {0x86feb0?, 0xc00015a000?}, 0x6?)
	/usr/local/go/src/net/http/server.go:2938 +0x8e
net/http.(*conn).serve(0xc00011a1b0, {0x8704c0, 0xc0000a3f80})
	/usr/local/go/src/net/http/server.go:2009 +0x5f4
created by net/http.(*Server).Serve in goroutine 1
	/usr/local/go/src/net/http/server.go:3086 +0x5cb

This is caused because the list nodes api returns records like this:

{
    "AddedToClusterTime": null,
    "BrokerNodeInfo": null,
    "ControllerNodeInfo": {
        "Endpoints": [
            "c-10002.foo.xxxxxx.c7.kafka.us-east-1.amazonaws.com"
        ]
    },
    "InstanceType": null,
    "NodeARN": null,
    "NodeType": "CONTROLLER",
    "ZookeeperNodeInfo": null
}

which have a nil BrokerNodeInfo

This PR fixes this bug, and also adds these controller nodes to the target endpoints.

When JMX Exporter and Node Exporter are enabled on the cluster, these nodes only seem to be running JMX Exporter, so we are only adding this endpoint for these nodes.

[errm]

Ping @joshm91

[joshm91]

Thanks for this, @errm. I've set some time aside next week to give this repo a bit of attention so I'll get your change reviewed and merged then.

[errm]

Awesome! Thanks 🙇

[joshm91]

Thanks again for this! Merged and pushed to prometheus-msk-discovery:0.7.0 and prometheus-msk-discovery:latest.