Add utf8::bom and utf16::bom helper rules.

[samhocevar]

I find myself reimplementing a rule for the optional byte-order mark in every UTF-8 parser I write. The character is automatically added by many Windows-based text editors, so it makes sense to have it in pegtl::utf8 and pegtl::utf16.

[coveralls]

Coverage remained the same at 99.904% when pulling 6eb4d7c on lolengine:feature-unicode-bom into b47152c on taocpp:master.

[d-frey]

Thanks, bom seems to be sufficiently general and is a welcomed addition. I also added a bom rule for UTF-32 to detect them, plus some cleanups.