If the entire input is stripped out, this used to return `nil` which
caused downstream parsing to fail. Instead, return `[]` which is the
Floki representation of an empty tree.
Fixes#36
Standard tags are returned by Mochiweb as binaries. The atom tags are
for special case parsing (such as php includes). Since that's not oging
to be part of the article, simply exclude those while normalizing.
Fixes#30
See also:
Mochiweb parser: 9608d786ef/src/mochiweb_html.erl (L345)
added fix for content-type with charset
updated function names to match elixir naming conventions (is_ vs ?)
minor version bump
added default content-type of text/plain when header is missing
