Editorial meeting consensus: Do not repeat a parent heading item in subordinate headings.

@jensmaurer: What about [container.node]?

@tkoeppe: [container.node] is odd. It specifies an exposition-only class node_handle, which need not exist under this name. Everywhere else in the vicinity, we specify requirements on classes or class templates by specifying the valid operations on them, instead of showing a class synopsis.

I'd like to point out that the superordinate heading says "Node handles" and the subordinate headings say "node_handle", which is not exactly a repetition (although quite close).

@jensmaurer: Yeah, that's true. Not sure what to do here. @zygoloid: Opinions?

My primary goal here is to get rid of a bold-italic-teletype font requirement, which we cannot satisfy :-)

I'm happy with removing node_handle from these headings, as they are not really describing members of some class named node_handle; rather, they're describing members of the exposition-only notion of a "node handle". Done in 2fcacf3.

Do you have a stance on the repetition in the headers' synopsis? That repetition might provide some context that's implicitly available in the subclauses for being nested.

@JohelEGP, I don't understand the question. Header synopses usually say "Header <blah> synopsis", and that doesn't mention a class name.

I mean the line comments like this:

    // [time.duration], class template duration
    template<class Rep, class Period = ratio<1>> class duration;

@JohelEGP, those comments quote the subclause heading, so those seem fine.

This should be complete now.