[finagle-memcached] Minor perf optimizations

Removed unnecessary toSeq, also made FrameDecoder an abstract class to avoid more expensive interface calls.

Differential Revision: https://phabricator.twitter.biz/D1173582

Author: mbezoyan

finagle-memcached/src/main/scala/com/twitter/finagle/memcached/protocol/text/FrameDecoder.scala

@@ -3,7 +3,7 @@ package com.twitter.finagle.memcached.protocol.text
 import com.twitter.io.Buf
 import scala.collection.mutable
 
-private[memcached] trait FrameDecoder[Result] {
+private[memcached] abstract class FrameDecoder[Result] {
 
   /**
    * Return the number of raw bytes needed, or -1 if a text line is needed.

finagle-memcached/src/main/scala/com/twitter/finagle/memcached/util/ParserUtils.scala

@@ -44,7 +44,7 @@ object ParserUtils {
         segmentStart = segmentEnd + 1
       }
     }
-    split.toSeq
+    split
   }
 
   /**
@@ -69,5 +69,4 @@ object ParserUtils {
       num
     }
   }
-
 }