Add smart detection of input language

[hackonteur]

I would like for coderay to be able to detect the input format by looking at the contents of the file. This would be particularly helpful in conjunction with #110: Add support for input from STDIN; since coderay can't use the file extension in that case.

I have two ideas for how to support this initially:

• Detect a unified diff style input, for the example of piping a diff to coderay:

diff -U1 foo.txt bar.txt | coderay

• Add support for Emacs mode hints. So coderay could understand that this means it should treat the input as C:

/* -*- mode: c -*- */

[korny]

I'm trying to understand how Pygments implements this. They seem to send the input to every possible scanner to check and rate it: https://bitbucket.org/birkenfeld/pygments-main/src/157c9feaccb86ac52da3d6b234cad92c85ee419c/pygments/lexers/agile.py?at=default#cl-1008

What do you think about this approach?