=================
* Important fixes for incorrect matches of the alphanumeric encoding mode.
Previously, the pattern would match if a single line was alphanumeric only
(even if others wern't). Also, the two characters ``{`` and ``}`` had snuck
in as valid characters. Thanks to Eran Tromer for the report and fix.
* Optimized chunking -- if the parts of the data stream can be encoded more
efficiently, the data will be split into chunks of the most efficient modes.