Fix build with GHC 7.3+

On chunk boundaries, the character was not recorded and the global index not
incremented, causing issue #10.

On a chunk boundary, we were not passing the correct mask and skip values
along to the function that would process the next chunk.

The "real" tests remain in tests/tests - this test suite is built without
optimization, and simply lets us do a quick pass/fail during automated builds.

Even though the value behind a Size is an Int, we actually intend that those
values should always be non-negative. (We don't use the notionally more
appropriate Word because GHC doesn't do a very good job with it.)

But non-negative means that 0+0 should be 0! Um, oops.

This code is derived from Björn Höhrmann's UTF-8 decoder.  Compared
to the original Haskell decoder from cac7dbcbc392, it's between
2.17 and 3.68 times faster.  It's even between 1.18 and 3.58 times
faster than the improved Haskell decoder from 71ead801296a.

The x86-specific decoding path gives a substantial win for entirely
and partly ASCII text, e.g. HTML and XML, at the cost of being about
17% slower than the portable C decoder for entirely non-ASCII text.