And for this reason alone and others explained here, I would suggest making the place as a node always the favored solution. The place as an area would be always an acceptable intermediate solution.
As I said I have no problem for the landuse=residential or whatever to include them, especially when mixed use and even vertically separated. But be careful of “planned community” vs gated community. A gated community can be relatively well defined by gated perimeter.
It makes sense, and I agree that, in general, a settlement place=* tagged as an area should not be rendered for the reasons given. On the other hand, I agree with the comment that doing this for place=neighbourhood would persuade mappers to “tag for renderer” by adding a landuse tag to the area. We can also say that this is the OSM-Carto’s intention, while other renderers such as OsmAnd and Mapbox do not.
Tagging it all as a node would probably lead to data loss. Knowing what neighbourhood a specific building is located in is useful information, as is a building within a large amenity=* area. This would not be a problem if a neighbourhood is equivalent to a boundary=administrative, because the administrative boundary would take this duty. But I’m not clear how we should handle an estate area that has nothing to do with the government. Mapping it similarly to amenity=* could work well.
I agree with the wiki that place=neighbourhood should be tagged as an area only when its boundary is obvious and there is no conflict. This solves the OSM-Carto’s “what the boundary means” problem. By the way, leaving the general neighbourhood as it is is fine, but we need a specific tag for the estates, such as your suggestion neighbourhood=gated, so that the OSM-Carto can render it properly. This approach would also allow the renderer to render each type of neighbourhood differently. At least, the residential and industrial neighbourhoods should not be rendered in the same way.
It’s not entirely good. Use of place= is affected by an estate’s scale of development relative to the surroundings, and much more local-dependent usage of place= requiring more adjustments. Adopting =neighborhood causes orthogonality with its other meanings, even when you use neighborhood=. In larger gated communities, there may be subdivisions/sections/parts that is conceptually equivalent to the =neighborhood level or lower. Using neighborhood= doesn’t extend to them, although it’s is a workable interim solution.
So there’s no single reliable indicator of a gated community or residential/industrial estate by using place=. While place= can be overlapping as a vague concept, this is opposite to a gated community or residential/industrial estate having well defined limits. There’s no guarantee a place= area meets this criteria, as it is often misused.