Rename ingestor-prefixed connection metrics shared with collector

[mkeesey]

Problem

Two Prometheus metrics carry the ingestor subsystem in their names but are populated by both the ingestor and the collector:

• adxmon_ingestor_active_connections (gauge) — Go var metrics.IngestorActiveConnections
• adxmon_ingestor_dropped_connections_total (counter) — Go var metrics.IngestorDroppedConnectionsTotal

Both are incremented in pkg/limiter/listener.go, which is wired up by pkg/http.NewServer. That HTTP server is used by the collector as well as the ingestor, so both processes emit these metrics. The names are misleading and the collector's Status log line scans for them by their ingestor-prefixed names (see collector/metrics/service.go).

Proposal

Rename the metrics and variables to something process-neutral, e.g.:

• adxmon_server_active_connections
• adxmon_server_dropped_connections_total

Update the Go variable names (ServerActiveConnections, ServerDroppedConnectionsTotal) and the substring matches in collector/metrics/service.go and ingestor/metrics/service.go accordingly.

Compatibility

This is a breaking change for any dashboards or alerts that reference the existing metric names. We should:

1. Audit dashboards under dashboards/ and build/k8s/dashboards/ for references.
2. Coordinate the rename with any external consumers / Kusto ingestion mappings if applicable.

Context

Discovered while cleaning up the per-process Status log lines (branch cleanup/remove-dead-ingestor-samples-logging). Decided to leave the rename for a separate change since the existing metric is populated correctly — only the name is misleading.