docs(xtream): cite the measured evidence for the category language gate

The gate's rationale named hypothetical category shapes. On a real
catalog the four it actually turns away are VOD (5,245 movies), KIDS
(1,010), SHOW and WWE — without it the language select offers "VOD" and
"KIDS" as languages. Comments, doc and one spec case only.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

docs/architecture/vod-multi-source.md

@@ -223,11 +223,13 @@ per `(playlist_id, xtream_id)` in both query tiers — `char(31)` because the
 default `,` appears inside real category names), and
 `unambiguousCategoryLanguage` reduces them: categories without a recognized
 language prefix abstain, all prefixed ones must agree, and a conflict yields
-nothing. Category prefixes must additionally pass `isKnownLanguageTag`,
-because categories routinely start with "NEW |", "TOP |" or "VIP |" — and
-`new`, `top` and `hot` are even assigned ISO 639-3 codes, which is why the
-gate is an `Intl.DisplayNames` check for two-letter codes plus a curated list
-for longer tags rather than a registry lookup. Only the pipe title form stays
+nothing. Category prefixes must additionally pass `isKnownLanguageTag`.
+Measured on a real catalog, that gate is what keeps "VOD" (5,245 movies),
+"KIDS" (1,010), "SHOW" and "WWE" out of a list of LANGUAGES; "NEW |",
+"TOP |" and "VIP |" are everyday shapes too, and `new`, `top` and `hot` are
+even assigned ISO 639-3 codes — which is why the gate is an
+`Intl.DisplayNames` check for two-letter codes plus a curated list for
+longer tags rather than a registry lookup. Only the pipe title form stays
 permissive: a tag before a pipe in a movie title is overwhelmingly a
 language, and tightening the legacy form would drop filter options that work
 today. The route's own row reads the one category the route arrived through

libs/shared/interfaces/src/lib/vod-source-language.util.spec.ts

@@ -95,6 +95,14 @@ describe('isKnownLanguageTag', () => {
         expect(isKnownLanguageTag('MULTI')).toBe(true);
     });
 
+    it('rejects the content-type prefixes categories actually use', () => {
+        // The four the gate turns away most often on a real catalog; without
+        // it the language select offers "VOD" and "KIDS".
+        for (const tag of ['VOD', 'KIDS', 'SHOW', 'WWE']) {
+            expect(isKnownLanguageTag(tag)).toBe(false);
+        }
+    });
+
     it('rejects everyday category words that are real ISO 639-3 codes', () => {
         // `new`, `top` and `hot` are assigned in ISO 639-3, which is exactly
         // why validation is curated instead of registry-driven.

libs/shared/interfaces/src/lib/vod-source-language.util.ts

@@ -20,11 +20,13 @@ import { PROVIDER_PIPE_CLASS } from './title-normalization.util';
  *   rip tags — "[HD] Dune", "NEW - Dune" — which would not only pollute the
  *   select but, since a title prefix outranks the category language, mask a
  *   real one and get the row excluded by the very filter meant to find it.
- * - Category names are noisiest: "NEW | 2024", "TOP | 250" and "VIP | Cinema"
- *   are everyday category shapes, and `new`, `top` and `hot` are even real
- *   ISO 639-3 codes, so a prefix read off a category always passes
- *   `isKnownLanguageTag`. Empty beats wrong: an unrecognized tag yields no
- *   language rather than a wrong filter option.
+ * - Category names are noisiest, so a prefix read off a category always
+ *   passes `isKnownLanguageTag`. Measured on a real catalog, the gate is what
+ *   keeps "VOD" (5,245 movies), "KIDS" (1,010), "SHOW" and "WWE" out of a
+ *   list of LANGUAGES; "NEW | 2024" and "TOP | 250" are everyday shapes too,
+ *   and `new`, `top` and `hot` are even assigned ISO 639-3 codes — which is
+ *   why the gate is curated rather than a registry lookup. Empty beats wrong:
+ *   an unrecognized tag yields no language rather than a wrong filter option.
  */
 
 /**