Take shadow DOM into account

[frivoal]

The spec does not currently say anything about shadow DOM. It most certainly should say something.

[jihyerish]

I think there are several things that we need to discuss shadow DOM.

1. How the focus navigates for the open shadow DOM whose shadow host has "tabindex=-1"?

In this case, seqnav works that the focus doesn't go inside the shadow DOM even if there is any focusable element within it.
Should the spatnav behave the same with seqnav?
I think spatnav would behave the same with seqnav which is the focus cannot move inside the shadow dom if the shadow host is specified with "tabindex=-1".

Depends on the discussion in #69, this behavior may be described in other way.
If shadow dom becomes spatnav container by default, it causes the conflict with the suggestion above.
Because the focus can move inside the spatnav container even if the spatnav container is specified with "tabindex=-1".

2. How the focus navigates for the closed shadow DOM?

It would be hard to know about which element has the focus within the shadow DOM.
(e.g. shadowRoot.activeElement returns null)
Can the browser know wheather the focus can move more inside the shadow dom?
Can the focus move out from the shadow dom if there isn't any focusable element?
Can navnotarget event be used for shadow DOM in this case?

[jihyerish]

In this case, seqnav works that the focus doesn't go inside the shadow DOM even if there is any focusable element within it.

There was a discussion about setting tabindex=-1 on a shadow host makes the elements inside the shadow DOM be skipped for the sequential navigation (tab navigation).
related issue: WICG/webcomponents#520 (comment)

[frivoal]

Issue migrated to w3c/csswg-drafts#3381