brontide+peer: use internal sync/pool to reduce allocations

This ensures that under medium to high load, we eliminate all
allocations once we arrive a steady state, re working memory.

peer/brontide.go

@@ -19,6 +19,7 @@ import (
 	"github.com/btcsuite/btcd/txscript"
 	"github.com/btcsuite/btcd/wire"
 	"github.com/btcsuite/btclog/v2"
+	"github.com/lightningnetwork/lnd/brontide"
 	"github.com/lightningnetwork/lnd/buffer"
 	"github.com/lightningnetwork/lnd/chainntnfs"
 	"github.com/lightningnetwork/lnd/channeldb"
@@ -2705,6 +2706,13 @@ out:
 				goto retry
 			}
 
+			// Message has either been successfully sent or an
+			// unrecoverable error occurred.  Either way, we can
+			// free the memory used to store the message.
+			if bConn, ok := p.cfg.Conn.(*brontide.Conn); ok {
+				bConn.ClearPendingSend()
+			}
+
 			// The write succeeded, reset the idle timer to prevent
 			// us from disconnecting the peer.
 			if !idleTimer.Stop() {