Fix record counts.

previously the record count could be wrong and reflect how much data the
worker was writting, independantly of the partition

kafka-connect-s3/src/main/java/io/confluent/connect/s3/TopicPartitionWriter.java

@@ -683,7 +683,7 @@ private void callbackFile(String encodedPartition) {
     fileCallback.ifPresent(fs -> fs.call(tp.topic(), encodedPartition,
             commitFiles.get(encodedPartition), tp.partition(),
             new DateTime(baseRecordTimestamp).withZone(timeZone),
-            new DateTime(currentTimestamp).withZone(timeZone), recordCount,
+            new DateTime(currentTimestamp).withZone(timeZone), recordCounts.get(encodedPartition),
             new DateTime(time.milliseconds()).withZone(timeZone)));
   }