refactor: split the repo into self-contained validator, directory, quickstart, spec and site folders

validator/ holds mewslint (lint, css, fetch, DTD/CSS data); directory/ holds
the mews.page service (app, db, check, pages, cli) and depends on validator/;
quickstart/ holds mews-quickstart and md2mews.py for third parties who want
to create Mews content without cloning the repo; spec/ consolidates doc/ and
dtd/ with the canonical mews-0.1.css; site/ becomes a Pelican project with
the forward-facing pages as Markdown, the check form kept as a raw HTML
snippet. Root pyproject.toml is now a virtual uv workspace over validator,
directory and quickstart, with cross-cutting tests that span more than one
of them staying at the repo root. Makefile and flake.nix updated to match.

site/pelicanconf.py

@@ -0,0 +1,44 @@
+AUTHOR = "randogoth"
+SITENAME = "the mews"
+SITEURL = ""
+PATH = "content"
+TIMEZONE = "UTC"
+DEFAULT_LANG = "en"
+THEME = "theme"
+
+# The plugin reads the allowed elements and attributes from the Mews DTD and
+# brings every page inside that subset.
+PLUGINS = ["pelican.plugins.mews"]
+
+# This whole site is Mews, so its pages are plain .html, and index.html is
+# the directory index everywhere. The CSS dir is left empty so the plugin
+# writes and links mews-0.1.css at the site root, where the spec's examples
+# and every other Mews page expect to find it.
+MEWS_SUFFIX = ".html"
+MEWS_CSS_DIR = ""
+# mews.page is the reference deployment: a page that doesn't conform must
+# fail the build, not ship with the offending markup quietly stripped.
+MEWS_STRICT = True
+
+PAGE_URL = PAGE_SAVE_AS = "{slug}.html"
+
+# Markdown has to produce XML for the plugin to parse. Tables and definition
+# lists are both in the Mews subset but need their extensions turned on.
+MARKDOWN = {
+    "output_format": "xhtml",
+    "extension_configs": {
+        "markdown.extensions.tables": {},
+        "markdown.extensions.def_list": {},
+    },
+}
+
+# This site is two hand-written pages, not a blog: no article stream, so none
+# of Pelican's own listing templates are needed.
+DIRECT_TEMPLATES = []
+AUTHOR_SAVE_AS = CATEGORY_SAVE_AS = TAG_SAVE_AS = ""
+AUTHORS_SAVE_AS = CATEGORIES_SAVE_AS = TAGS_SAVE_AS = ARCHIVES_SAVE_AS = ""
+FEED_ALL_ATOM = FEED_ALL_RSS = None
+CATEGORY_FEED_ATOM = CATEGORY_FEED_RSS = None
+AUTHOR_FEED_ATOM = AUTHOR_FEED_RSS = None
+TAG_FEED_ATOM = TAG_FEED_RSS = None
+TRANSLATION_FEED_ATOM = TRANSLATION_FEED_RSS = None