Fix metric names: strip loki_process_custom_ prefix, drop internal labels

Alloy's stage.metrics prefixes all metric names with
loki_process_custom_. Add a relabel rule to strip the prefix so
dashboards can query clean names (flyio_nginx_http_requests_total
etc). Also drop component_id/component_path/filename labels.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

fly/alloy.river

@@ -110,10 +110,24 @@ prometheus.scrape "self" {
   scrape_interval = "15s"
 }
 
-// Add instance label to all metrics
+// Strip the "loki_process_custom_" prefix that Alloy adds to stage.metrics,
+// then add instance label. This keeps dashboard queries clean.
 prometheus.relabel "instance" {
   forward_to = [prometheus.remote_write.prometheus.receiver]
 
+  rule {
+    source_labels = ["__name__"]
+    regex         = "loki_process_custom_(.*)"
+    target_label  = "__name__"
+    replacement   = "$1"
+  }
+
+  // Drop internal labels added by the loki pipeline
+  rule {
+    regex  = "component_id|component_path|filename"
+    action = "labeldrop"
+  }
+
   rule {
     target_label = "instance"
     replacement  = "flyio-proxy"