[expr.const] p5 integrate construct_at/destroy_at wording into bullet 2 in p4

[jensmaurer]

CWG in Belfast: This needs a CWG nod once there is a specific suggestion.

[frederick-vs-ja]

This seems resolved by P2747R2 (#7097). Now construct_at no longer needs magic.

[jwakely]

Yup, the text about construct_at and destroy_at is no longer present in [expr.const]. I'll let Jens close this though, since it's his issue.