Add metrics for out of view statement distribution errors (#4972)

* Add a simple metric for statements out-of-view

* Avoid repeated out-of-view peer reputation change messages

* Log reporting status

* Address review comments

* Use counter to store a number of unexpected messages from a peer

* Distinguish different unexpected statements in the metrics

* Fix labels cardinality

* Rename metric name to `statements_unexpected`

* Move metrics to a separate unit, avoid unnecessary enum

* Prefer specific methods in lieu of public constants

polkadot/node/network/statement-distribution/src/requester.rs

@@ -32,7 +32,7 @@ use polkadot_node_subsystem_util::TimeoutExt;
 use polkadot_primitives::v1::{CandidateHash, CommittedCandidateReceipt, Hash};
 use polkadot_subsystem::{Span, Stage};
 
-use crate::{Metrics, COST_WRONG_HASH, LOG_TARGET};
+use crate::{metrics::Metrics, COST_WRONG_HASH, LOG_TARGET};
 
 // In case we failed fetching from our known peers, how long we should wait before attempting a
 // retry, even though we have not yet discovered any new peers. Or in other words how long to
@@ -117,6 +117,7 @@ pub async fn fetch(
 				Ok(StatementFetchingResponse::Statement(statement)) => {
 					if statement.hash() != candidate_hash {
 						metrics.on_received_response(false);
+						metrics.on_unexpected_statement_large();
 
 						if let Err(err) =
 							sender.feed(RequesterMessage::ReportPeer(peer, COST_WRONG_HASH)).await
@@ -161,6 +162,7 @@ pub async fn fetch(
 					);
 
 					metrics.on_received_response(false);
+					metrics.on_unexpected_statement_large();
 				},
 			}