Merge pull request gabriel-vasile#188 from gabriel-vasile/utf-control…

…-chars

Exclude weird control characters from valid utf8 text/plain

Check for EOF in DetectReader; for gabriel-vasile#162

Before v1.2.0, DetectReader returned text/plain for empty inputs.
After v1.2.0, it started returning EOF error.
This commit changes DetectReader behavior to how it was before v1.2.0

Make readme shorter

Enable race detector in tests

Docs improvements

Refactor mso xml detection for trailing signatures

Zip archives do not care about the order of the files inside them. This
makes the signature files of Ms office formats to be in the trailing
part of the archive and not get picked up by the detection alg. This
commit improves detection by giving up search for general MS office
signature search in favor of searching for more precise entries.
Related issue: gabriel-vasile#81

Add docs for EqualsAny

Improve zip sub formats detection

Add Performance section in README

Use Is method in tests instead of string comparƣson

Behave as http.DetectContentType and return text/plain for empty input