Ensure exported NATS Server metrics have the correct TYPE information for the origin data

[bwerthmann]

Presently, all metrics exported matching metric name nats_core* have TYPE gauge. A subset of these are actually other types of metrics.

The distinction between counter and gauge is significant. On the consumer side, different approaches are needed in PromQL.

Example counter metric: nats_core_total_connection_count

This is a counter because it 'increases forever' (until the process restarts).

# HELP nats_core_total_connection_count Total number of client connections serviced gauge
# TYPE nats_core_total_connection_count gauge
nats_core_total_connection_count{server_cluster="...",server_id="...",server_name="..."} 6345789

Example gaugue metric: nats_core_connection_count

# HELP nats_core_connection_count Current number of client connections gauge
# TYPE nats_core_connection_count gauge
nats_core_connection_count{server_cluster="...",server_id="...",server_name="..."} 1

NOTE: Some non-gauge metrics are present in the scrape

There are all associated with the surveyor process, not any NATS Server.

# TYPE go_gc_duration_seconds summary
# TYPE go_memstats_alloc_bytes_total counter
# TYPE go_memstats_frees_total counter
# TYPE go_memstats_lookups_total counter
# TYPE go_memstats_mallocs_total counter
# TYPE nats_survey_duration_seconds summary
# TYPE nats_survey_late_replies_count counter
# TYPE nats_survey_nats_reconnects counter
# TYPE nats_survey_no_replies_count counter
# TYPE process_cpu_seconds_total counter
# TYPE promhttp_metric_handler_requests_total counter