fix(web): one language chip per file, not per track

Commentary tracks carry the same language as the feature audio, so a file
row rendered `en` two or three times. The chips answer what a file can be
watched in; repeats say nothing the first said.

web/src/releases.ts

@@ -357,6 +357,8 @@ export function bucketOf(release: MovieRelease): Bucket {
  * The probed attributes a file row shows as chips (§5.6, §7.4) — ffprobe
  * truth, not name claims. Shared by movie and episode file rows; the probe
  * JSON arrives untyped from the API, so the shape is narrowed here.
+ *
+ * Audio is reported as the set of languages present, one chip each.
  */
 export function probedAttributeTags(probed: unknown): string[] {
   if (probed === null || typeof probed !== "object") {
@@ -378,11 +380,15 @@ export function probedAttributeTags(probed: unknown): string[] {
   if (probe.hdr && probe.hdr !== "SDR") {
     tags.push(probe.hdr);
   }
+  // Set, not push: commentary tracks repeat the feature audio's language,
+  // and no test catches a row that renders `en` three times.
+  const languages = new Set<string>();
   for (const track of probe.audio_tracks ?? []) {
     if (track.language) {
-      tags.push(track.language);
+      languages.add(track.language);
     }
   }
+  tags.push(...languages);
   return tags;
 }