Refactor sanitize_token for better subclassing

[mfa]

For my application I need to clean all unallowed tokens.
At the moment the only way is to copy the whole sanitize_token method, remove a few lines and add one.
Splitting up the sanitize_token method would simplify this a lot and reduces code duplication (in my code).

Example usage:

import html5lib
from html5lib import treebuilders, treewalkers, serializer, sanitizer

class MySanitizer(sanitizer.HTMLSanitizer):
    # reduce tokens to only a few
    acceptable_elements = ['b', 'br', 'center', 'em', 'h3', 'h4', 'h5', 'h6', 
                           'i', 'li', 'ol', 'p', 'span', 'strike', 'strong', 
                           'tt', 'ul']

    allowed_elements = acceptable_elements

    def unallowed_token(self, token):
        # remove all unallowed tokens
        return ""

def sanitize(input):
    p = html5lib.HTMLParser(tokenizer=MySanitizer, tree=treebuilders.getTreeBuilder("dom"))
    dom_tree = p.parseFragment(input)
    walker = treewalkers.getTreeWalker("dom")
    stream = walker(dom_tree)

    s = serializer.htmlserializer.HTMLSerializer(omit_optional_tags=False,
                                                 quote_attr_values=True)
    return u"".join(s.serialize(stream))

[gsnedders]

Well, Travis CI shows you having made all the sanitizer tests fail. That ought to be fixed. Otherwise, you should use the standard English "disallow" and not "unallow". That said, in principle looks good.

[gsnedders]

Merged in 52f9ca6