Prometheus metrics should be grouped by type #1045
Description
Describe the bug
Output from prometheus endpoints should be grouped by type not by resource.
For example
# TYPE lavinmq_detailed_queue_messages_ready gauge
# HELP lavinmq_detailed_queue_messages_ready Messages ready to be delivered to consumers
lavinmq_detailed_queue_messages_ready{queue="queue_1", vhost="cloudamqp"} 0
lavinmq_detailed_queue_messages_ready{queue="queue_2", vhost="cloudamqp"} 0
But now LavinMQ does it like this
# TYPE lavinmq_detailed_queue_messages_ready gauge
# HELP lavinmq_detailed_queue_messages_ready Messages ready to be delivered to consumers
lavinmq_detailed_queue_messages_ready{queue="queue_1", vhost="cloudamqp"} 0
# TYPE lavinmq_detailed_queue_messages_ready gauge
# HELP lavinmq_detailed_queue_messages_ready Messages ready to be delivered to consumers
lavinmq_detailed_queue_messages_ready{queue="queue_2", vhost="cloudamqp"} 0
Reference: https://prometheus.io/docs/instrumenting/exposition_formats/
Not only is this correct but it will also reduce the number bytes going out on the wire for each request.