Extend the prolog to support arbitrary metadata

[ndw]

I sent email about supporting grammar metadata a while back. That discussion petered out on the mailing list, but I still think it's worth persuing, so here's an issue about it.

The salient bit is:

Grammar metadata

The prolog may also contain additional metadata about the grammar.

 metadata: name, (field++msep | RWS, chardata), whitespace*, -".", RS .
    field: RWS, name, whitespace*, -":", whitespace*, chardata .
-chardata: -'"', -dchar+, -'"';
           -"'", -schar+, -"'".
    -msep: whitespace*, -"," .

Each metadata declaration consists of a name followed by either a string or a set of fields. Each field value is a string.

For example:

ixml version "1.1" .
author  givenname: "Norman", surname: "Tovey-Walsh" .
license id: "MIT", name: "The MIT License",
        uri: "https://opensource.org/license/mit" .
date    "2024-03-12" .
release "1.0" .

…

As a nod towards some level of standardization, I believe this is sufficient to capture the citation metadata suggested for project repositories.

[ndw]

A suggestion was made that the grammar and the test data are out-of-sync.

[ndw]

Also: is chardata needed?