Add encoding functions

[shaneaevans]

These are loosely based on the encoding in scrapy.

Main differences:

• tweaks to regular expressions for encoding detection in HTML. One regexp handles html and xml
• handle byte order marks
• better handling of character encoding overrides, with an updated list
• does not fall back to BeautifulSoup, instead the auto-detect is customizeable and disabled by default

[dangra]

Are python byte slices cheap?

Just to point about rest_of_data been copied and later dropped when bom_enc and enc doens't match.
It's a rare case I know, but it waste memory for big responses or for sites sending BOM and different transport encoding for all its pages.

I think this isn't a merge blocker but worth pointing it, once in the wild we can optimize if affects any real case.

[shaneaevans]

It should be pretty rare that we have BOMs that differ from the encoding in the headers, and it's not that expensive. But still - why take the risk? I'll change it.