Arithmetic coding is a feature of the JPEG standard. Although libjpeg
had always implemented arithmetic coding, it was compiled out by default
due to patents.

Those patents have now expired.

libjpeg 8 now enables arithmetic coding by default. Distributions which
use libjpeg < 8 can also support arithmetic coding by using the
appropriate CFLAGS to enable it. libjpeg-turbo 1.3.1 also has support
for arithmetic coding and is in popular shipping distributions.

Software such as jpegtran can losslessly convert Huffman compressed
images to arithmetic coding and vice versa. The lossy behavior of JPEG
does not happen at this (bit coding) layer of the format.

This initial patch provides a checkbox (disabled by default) to create
files which use arithmetic coding. It also has a tooltip warning that
such files may not be compatible with older decoders.