Undo optimization

In Hive coercion tests some test cases produce binary arrays that contain \0.

Previously we were writing Slices representing UTF-8 String by calling .toStringUtf8()
which passed the bytes to the String constructor that would truncate the value at \0.

When directly writing Slice to a JSON output, \0 isn't interpreted as String terminator,
got escaped and appears in the output without truncation.

It seems more correct but breaks previous behaviour.

core/trino-main/src/main/java/io/trino/server/protocol/JsonEncodingUtils.java

@@ -263,8 +263,7 @@ public void encode(JsonGenerator generator, ConnectorSession session, Block bloc
                 return;
             }
             Slice slice = VARCHAR.getSlice(block, position);
-            // Optimization: avoid conversion from Slice to String and String to bytes when writing UTF-8 strings
-            generator.writeUTF8String(slice.byteArray(), slice.byteArrayOffset(), slice.length());
+            generator.writeString(slice.toStringUtf8());
         }
     }
 
@@ -287,8 +286,7 @@ public void encode(JsonGenerator generator, ConnectorSession session, Block bloc
                 return;
             }
             Slice slice = padSpaces(VARCHAR.getSlice(block, position), length);
-            // Optimization: avoid conversion from Slice to String and String to bytes when writing UTF-8 strings
-            generator.writeUTF8String(slice.byteArray(), slice.byteArrayOffset(), slice.length());
+            generator.writeString(slice.toStringUtf8());
         }
     }