docs: record that CI, the registry and the deploy moved to Gitea

The pipeline change made a set of sentences false, and they are the only
record of why things are the way they are, so they move with it: the CI
section and deploy target in README, the cron and BASE_PATH notes in
ARCHITECTURE, the documented User-Agent in INGESTION, and the contact
URL in SOURCES.

Two comments in source were false in the same way and are fixed here
rather than left for the next reader. `serve.ts` said compression was
"the whole difference between the Docker image and the deployed site" —
the image *is* the deployed site now, so that negotiation is the only
compression any reader gets. `robots.ts` justified product-token
matching with "our contact URL contains the string StereotypicalCat",
which it no longer does; the hazard is the arbitrary substring, not that
one, so the comment now says so and the regression test keeps the URL
that actually caused it.

What is deliberately NOT rewritten: game8.co's refusal stays in the past
tense with the Gitea result marked unverified, in AGENTS.md, README,
SOURCES and FEEDBACK alike. A new runner address is the variable that
failure turns on, so it may well be served — but "may well be" is not
evidence, and the four places that would have to change are named
together so they change together once a cycle has actually run.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>

AGENTS.md

@@ -97,7 +97,7 @@ bun run parse endfield-wikigg-events fixtures/endfield/wikigg-events-2026-08-15.
 bun test test/dates.test.ts
 bun test --test-name-pattern "year-less"
 
-# Hosting under a subpath (GitHub Pages)
+# Hosting under a subpath (CI does not set this — the image serves from /)
 BASE_PATH=/gacha-event-tracker/ bun run build
 ```
 
@@ -348,11 +348,11 @@ AI-training crawlers. Our use is a low-rate personal aggregator with attribution
 training, and no `User-agent: *` rule applies to our paths. Keep it that way: do not raise the fetch
 rate, and do not add an LLM that consumes page content.
 
-**game8.co does not answer a GitHub Actions runner** (confirmed 2026-08-17). Its edge returns
+**game8.co did not answer a GitHub Actions runner** (confirmed 2026-08-17). Its edge returned
 `202 Accepted` with a bot-management body to every one of the nine game8 sources, from the first
 scheduled cycle onward — `last confirmed: never` — while the same URLs return `200` and parse
-cleanly from a normal address. So `robots.txt` permits us and the network does not, and those nine
-games have only ever been built from checked-in fixtures in CI.
+cleanly from a normal address. So `robots.txt` permitted us and the network did not, and those nine
+games had only ever been built from checked-in fixtures in CI.
 
 The per-host spacing above does not fix this and was not meant to: a 202 on the very first request
 of a cycle is address reputation, not rate. **Do not work around it.** Browser-shaped headers, a
@@ -361,6 +361,16 @@ same reason `uma.moe` was declined below — and unlike `uma.moe` we would be do
 `robots.txt` was welcoming, which makes it worse, not better. The legitimate options are to run the
 refresh from an address game8 will serve, or to find those games another source.
 
+**CI moved to a Gitea runner on 2026-09-16, which is the first of those two options — and its result
+is not yet known.** A different address is exactly the variable this failure turns on, so the nine
+game8 sources may now refresh on schedule, or may not. Nothing here should be rewritten on the
+strength of that hope: the tense above stays past because the GitHub evidence is real and the Gitea
+evidence does not exist yet. Establish it the same way it was established before — let a cycle run,
+then read `git log` on `snapshots/` for which hosts actually landed — and only then update this
+section, `README.md` § Status, `docs/SOURCES.md` § Method and `docs/FEEDBACK.md`'s P0 row together.
+A single manual `--only` dispatch is enough to answer it for one source; note that `--dry-run`
+cannot, because it makes no requests at all by design.
+
 A source whose ToS forbids automated access does not get an adapter. Flag it and ask.
 
 **Sources assessed and declined** (2026-08-17, extended 2026-08-19), so these are not