Clarify how `<script type=webbundle>` affects speculative HTML parsing.

[hayatoito]

We should clarify how <script type=webbundle> affects speculative HTML parsing.

A browser must not fetch a resource from the network speculatively if <script type=webbundle> says the resource should be fetched from the bundle.

Regarding a browser's implementation, the chromium already implements this kind of behaviour in html_preload_scanner, as far as I understand correctly.

Note that this was pointed out here. Much appreciated!

[hayatoito]

In addition to that, HTMLPreoadScanner would need to now be able to read JSON in order to know how to preload the bundle.