implementation of Buffer.byteLength at fromListPartial

A continued effort to ensure the highWaterMark is measured against
byteLength in all instances. The buffer head data has been measured
with string.proto.length. This commit changes the method to byteLength.

lib/_stream_readable.js

@@ -856,7 +856,7 @@ function fromList(n, state) {
     state.buffer.clear();
   } else {
     // read part of list
-    ret = fromListPartial(n, state.buffer, state.decoder);
+    ret = fromListPartial(n, state.buffer, state.encoding, state.decoder);
   }
 
   return ret;
@@ -865,13 +865,13 @@ function fromList(n, state) {
 // Extracts only enough buffered data to satisfy the amount requested.
 // This function is designed to be inlinable, so please take care when making
 // changes to the function body.
-function fromListPartial(n, list, hasStrings) {
+function fromListPartial(n, list, encoding, hasStrings) {
   var ret;
-  if (n < list.head.data.length) {
+  if (n < Buffer.byteLength(list.head.data, encoding)) {
     // slice is the same for buffers and strings
     ret = list.head.data.slice(0, n);
     list.head.data = list.head.data.slice(n);
-  } else if (n === list.head.data.length) {
+  } else if (n === Buffer.byteLength(list.head.data, encoding)) {
     // first chunk is a perfect match
     ret = list.shift();
   } else {