Is collection subsetting too advanced / uncommon to be included in this spec?

[letmaik]

Section 8 has an example where a subsetted coverage belongs to the correspondingly subsetted collection. A subsetted collection means that all coverages in that collection are subsetted to some criteria. An example:

The EN3 dataset of the Met Office is a collection of many globally distributed vertical profile measurements (measurement of ocean temperature and salinity for several depths). A typical use case is to filter the whole collection for those vertical profiles within a given time frame and then subset those to the depth value closest to a given input (e.g. -20 metres) in order to compare those point values to a model grid.

Without collection subsetting, the filtered vertical profiles would have to be transmitted in full and subsetted locally, which may in some cases result in a too big data volume.

The question is, how advanced is this use case and the concept of collection subsetting (which I just use without having a clear referencable definition), and does it make sense to include it in the spec, as in the example response of section 8 (see the "inCollection" -> "subsetOf" path).