Change buffer file naming

[jfzunigac]

Changing the buffer file naming to logName.dirName.fileName.buffer, which follows the watermark file naming model. This helps by:

1. Reduce the amount of buffer files generated: the previous naming model created too many buffer files because of the uuid appended at the end.
2. Allow S3Writers to continue from buffer file after restart in case upload fails when closing the writer
3. Avoid multi-directory conflicts
4. Avoid conflicts between two or more different log streams processing the same file

Other conflicts can be handled by specifying a different s3 buffer directory per log stream

Also cleaning the classes by removing unused code and tests