This was written because I got sick of text I entered in many
applications having ugly "straight quotes" instead of the more
attractive “curly” quotes. I found it needlessly awkward to type the
Unicode characters necessary to do that, and it was easy enough to code
up a state machine that took care of most of the common cases one
encounters in English language texts.

While doing so, I also got the itch to experiment with using the Python
array module and the buffer protocol to implement writable strings,
because I was curious as to how much a fairly straightforward attempt at
same might speed up code (such as the state machine I had just written)
that makes many tiny, single-character modifications to a long strong.
It turns out that it is *very* hard to improve on Python's own strings,
but I left my own code in because its performance isn't all *that*
awful, either.