Optionaloptions: RenderNamesOptionsThe standalone form of the roster, and the one place grouping by kind is right. The workbook report carries a single defined-name section, deliberately unsplit: its job is triage, and a triage sort is only worth anything as one ordered reading. This report has the opposite job — "what exists, and what is it" — and a reader here has already chosen to look at names rather than at the book. So the split lands here and nowhere else.
What it holds is decided by where the collection came from. wb.getNames() is every named
item at any scope; sheet.getNames() is what is visible FROM that sheet, with shadowing already
resolved. Both render through this entry unchanged — the scope question was answered by the call
that produced the collection, which is exactly why there is no scope option here to get wrong.
Hidden names are included. Existing never hides (the same house rule that puts hidden sheets in the workbook roster), and this is the report that has to be complete: a hidden broken name with dependents is precisely the row a debugging agent must see, and the triage sort surfaces it.
The dependents column is the engine's answer, and it is read off COMPILED formulas, so the
caller settles the engine first — renderMarkdown settles its workbook argument before
dispatching here.
Renders the name manager — every named item in the collection, grouped by kind.