Support non-UTF8 encodings

[benogle]

Thanks to @aki77 we are better at handling utf8 with #25. But support for any other encodings is terrible.

• It assumes many non-utf-8 files are binary. Fix / get Assumes files with non-utf-8 encodings are binary gjtorikian/isBinaryFile#22 fixed
• Node only supports a few encodings. No bueno. We should use something like https://github.com/ashtuchkin/iconv-lite (which is what pathwatcher uses.
• Expose the encoding option to PathSearcher and PathReplacer

Potential files for testing: https://github.com/benogle/encodings

Open questions

• With open files, we can pass in the encoding from Atom. How to effectively search files with different encodings? I suppose we need to solve the encoding autodetection in core, then we can use the same pattern when opening each file.
• Same goes for replacing. We need to write the file in the correct encoding.