Shadow DOM elements are not seen by the script

[FremyCompany]

http://chromeexperiments.com should show more than 57 elements in Chrome, where the shadow dom works. It should continue to report 53 in Edge where nothing shows up.

[FremyCompany]

Investigating this further shows that solving this issue is analogous to be able to crawl content in iframes, security considerations excepted. We need to ask each element for its possible shadow dom, and if we find one we have to queue the execution of both the DOM and CSS analysis on that "document".

We can try to reuse the same CSSResults instance and accumulate the results between the runs, but we have to make sure that actually works. Right now some subsections of the data collection "erase" parts of the data structure while they create it. Supporting multi-document-roots runs would require updating this to make sure we only "add" or "increment" stuff, and that we never "set" result data.