Implement <select> parser “relaxation” — for “customizable” <select>

[sideshowbarker]

This change, in order to support the "customizable" <select> element:

• "relaxes" parsing rules in <select> (to allow more elements as children)
• adds the <selectedcontent> element for cloning selected <option> content
• aligns all select-related tree construction with the WHATWG spec (whatwg/html PR #10557)

ElementName.java

• Add <selectedcontent> element with TreeBuilder.SELECTEDCONTENT group

TreeBuilder.java

Structural changes:

• Remove dead IN_SELECT / IN_SELECT_IN_TABLE constants and all associated mode handlers and end-tag special cases — with select "relaxation", those modes are never entered
• Fix resetTheInsertionMode to skip past <select> elements on the stack instead of incorrectly returning IN_BODY — an enclosing <td>/<th> now correctly yields IN_CELL, an enclosing <table> yields IN_TABLE, etc.

Start-tag alignment with spec:

• <hr>: use generateImpliedEndTags() instead of manual isCurrent/pop
• <option>: remove redundant pop loop (spec says parse error only after generateImpliedEndTagsExceptFor("optgroup"))
• <optgroup>: remove redundant pop loop, add missing reconstructTheActiveFormattingElements(), fix parse error to check both <option> and <optgroup> in scope
• <select>: check fragment case first, remove generateImpliedEndTags from nested case, add framesetOk = false
• <input>: pop select and reprocess when select is in scope
• Table elements (<caption>, <tbody>, <tr>, <td>, <th>): treat as stray start tags, removing the old IN_SELECT_IN_TABLE dual-scope check

End-tag changes:

• Fold </select> into the standard block-element end-tag group (with </div>, </fieldset>, </button>, etc.)
• Remove dedicated </option> and </optgroup> end-tag cases — "any other end tag" handles them

Cloning hook:

• Add optionElementPopped() hook called from all pop() variants
• Default is no-op; DOM-side subclasses override to implement "maybe clone an option into selectedcontent"

Spec comments:

• Add spec URLs and verbatim spec-text comments throughout all select-related code sections

DOMTreeBuilder.java

• Implement optionElementPopped(): walk ancestors to find <select>, find <selectedcontent> descendant, check selectedness, deep-clone option children via cloneNode(true)

SAXTreeBuilder.java

• Implement optionElementPopped() as no-op (streaming/SAX mode has no live DOM to clone into)

XOMTreeBuilder.java

• Implement optionElementPopped() using XOM's Node.copy() for deep cloning

BrowserTreeBuilder.java

• Implement optionElementPopped() using cloneNodeDeep() for deep cloning

CharBufferNode.java

• Add getBuffer() accessor for direct char[] access during cloning

ParentNode.java

• Add clearChildren() method, to support clearing <selectedcontent>

[hsivonen]

Sorry about the delay.

As discussed on Matrix, I think the way this patch makes the cloning work on what the parser saw is incorrect and the spec instead requires the cloning to work on what's in the live DOM, which may be different if the option element opens, the network stalls, setTimeout modifies the DOM, and network resumes.

Further remarks in an inline comment.

[hsivonen]

I'm sorry that I've been in a very bad time management loop.

This looks very good, but there are some minor issues.

Thank you!

[sideshowbarker]

@hsivonen OK, all comments now resolved, I think. Ready for you to review again.

[hsivonen]

Unlike the other subclasses, this one does not appear to check the HTMLness of either select or selectedcontent. It's likely possible to fool at least the selectedcontent search with SVG. Can be a follow-up to fix.

[sideshowbarker]

I’ll take a shot at this before we merge

[hsivonen]

Let's merge this. Thank you and sorry about delay.