Reword description of rel decomposes to describe nesting vs aggregates accurately

Originally picked up by @hlg https://forums.buildingsmart.org/t/outdated-specification-semantics-of-ifcrelnests/2557

IfcRelDecomposes has this sentence:

> Decompositions may be constrained by requiring both, the whole and its parts, to be of the same type - thus establishing a nesting relationship. Or they may require some form of physical containment, thus establishing special types of aggregation relationships.

This sentence is outdated. The difference between aggregation and nesting is that aggregations are unordered, and nesting is ordered.

Here is a new proposed sentence to replace the above:

> Decompositions may be either an unordered aggregation of parts, or an ordered nested list of parts. A decomposition may also represent a physical decomposition of projections and voids which contribute to the shape of the element.

Then this note may be removed:

> NOTE There are two special names for decomposition, which are linguistically distinguished, nesting and aggregation. The subtypes of IfcRelDecomposes will introduce either the nesting or aggregation convention (see [IfcRelNests](http://localhost:8001/schema/ifckernel/lexical/ifcrelnests.htm) and [IfcRelAggregates](http://localhost:8001/schema/ifckernel/lexical/ifcrelaggregates.htm)).