Fix handling of uncompressed JAR entries

The Uint8Array returned from zip.read is just an offset into the jar's array
buffer. We need to properly use the byte length and offset when carving out a
slice.

classes.js

@@ -49,7 +49,7 @@ Classes.prototype.loadFile = function(fileName) {
         var zip = classfiles[name];
         if (fileName in zip.directory) {
             var bytes = zip.read(fileName);
-            data = bytes.buffer.slice(0, bytes.length);
+            data = bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength);
         }
         return !data;
     });