feat: URL normalisation, strict URLs and the one-hop redirect decision

- Normalize: https, www or bare host, double slashes and index files,
  trailing slash add/remove, optional ASCII lowercasing (defaults change
  nothing but scheme, slashes and index files)
- strict URLs through signpost MatchPath: a path that names a record
  without being its address 301s to the record's address
- resolve runs normalisation, redirects, strict URLs, then SEO; the
  location is the final address, so a visitor makes one hop
- P1 warns when a record's canonical URL redirects

controllers/api/resolve.go

@@ -13,11 +13,14 @@ import (
 var log = diary.For("seo")
 
 // Resolve serves GET /_seo/api/v1/resolve?url=<path or absolute URL>: the
-// effective SEO of a public URL for the frontend's <head>. It answers 200
-// with the resolve answer (action ok or not_found, the status the page
-// should answer with) and 422 when url is missing or not a path or an
-// http(s) URL. Absolute URLs come from the configured site URL, never from
-// the request's Host.
+// redirect decision and the effective SEO of a public URL for the
+// frontend (D-20a, D-103). It answers 200 with the resolve answer: action
+// redirect with status (301 or 302) and an absolute location, reached in
+// one hop through normalisation, redirects and strict URLs; gone with 410;
+// ok with the page's SEO; or not_found with 404, the status the page
+// should answer with. It answers 422 when url is missing or not a path or
+// an http(s) URL. Absolute URLs come from the configured site URL, never
+// from the request's Host. The frontend enforces the decision (Phase 33).
 func Resolve(service func() *classes.Service) http.HandlerFunc {
 	return func(w http.ResponseWriter, r *http.Request) {
 		svc := service()