Update tokenizer to match spec

This CL is a rough pass over the HTMLTokenizer to align it with parsing.md.
We'll need to do another pass more carefully in the future, but this CL gets us
roughly in the right ballpark.

We're not handling EOF properly. The parsing.md spec doesn't push the EOF
though the parser, which breaks our current way of handling EOF. We do ok if we
get EOF in the DataState, and that's enough to pass the tests for now.

Also, update camel-case.sky to reflect the fact that the parser doesn't
lower-case tag names anymore.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/678263002

engine/src/flutter/tests/lowlevel/camel-case.sky

@@ -4,7 +4,7 @@
 <script>
 describe('Element tag names', function() {
   it('should have various casing', function() {
-    assert.equal(document.documentElement.tagName, 'camelcase');
+    assert.equal(document.documentElement.tagName, 'CamelCase');
 
     var element = document.createElement('CamelCase');
     assert.equal(element.tagName, 'CamelCase');