Add dictation-locale transcription (incl. Russian), rework list capture UI

- TranscriptionSettings/Service route per-locale to SpeechTranscriber or
  DictationTranscriber (union of both supported sets, speech preferred);
  neutral-confidence ranking fallback keeps mixed-engine auto-pick working
- NotesListView: text + voice capture buttons above the list (glassProminent),
  search moved to the iOS 26 toolbar search accessory (.minimize)
- Changelog entries for languages, capture buttons, search placement

NotesTests/LanguagePickerTests.swift

@@ -36,6 +36,71 @@ struct LanguagePickerTests {
         #expect(LanguagePicker.meanConfidence(of: transcript([("no attributes", nil)])) == 0)
     }
 
+    // MARK: - Ranking confidence (mixed-engine graceful degradation)
+
+    @Test func hasConfidenceDetectsAttributedRuns() {
+        #expect(LanguagePicker.hasConfidence(of: transcript([("sure", 0.9)])))
+        #expect(LanguagePicker.hasConfidence(of: transcript([("a", nil), ("b", 0.2)])))
+        #expect(!LanguagePicker.hasConfidence(of: transcript([("none", nil)])))
+        #expect(!LanguagePicker.hasConfidence(of: AttributedString("")))
+    }
+
+    @Test func rankingConfidenceUsesRealMeanWhenPresent() {
+        // With any confidence attribute, ranking == the character-weighted mean.
+        let attributed = transcript([("hello", 0.9), (" world", 0.4)])
+        #expect(abs(LanguagePicker.rankingConfidence(of: attributed)
+            - LanguagePicker.meanConfidence(of: attributed)) < 0.0001)
+    }
+
+    @Test func rankingConfidenceFallsBackToNeutralForUnattributedText() {
+        // A DictationTranscriber transcript with text but no confidence attribute
+        // scores the documented neutral value, not 0, so it can still compete.
+        let dictation = transcript([("привет мир", nil)])
+        #expect(LanguagePicker.rankingConfidence(of: dictation) == LanguagePicker.neutralConfidence)
+    }
+
+    @Test func rankingConfidenceOfEmptyIsZero() {
+        #expect(LanguagePicker.rankingConfidence(of: AttributedString("")) == 0)
+    }
+
+    @Test func neutralDictationBeatsLowConfidenceSpeech() {
+        // Audio is Russian: the dictation locale produces good (neutral-scored)
+        // text; the mismatched speech locale produces low-confidence garbage.
+        let winner = LanguagePicker.pick(from: [
+            LanguageCandidate(localeID: "en-US", meanConfidence: 0.2, transcriptLength: 12),
+            LanguageCandidate(localeID: "ru-RU",
+                              meanConfidence: LanguagePicker.neutralConfidence,
+                              transcriptLength: 18),
+        ])
+        #expect(winner?.localeID == "ru-RU")
+    }
+
+    @Test func highConfidenceSpeechBeatsNeutralDictation() {
+        // Audio is English: the speech locale's real high confidence outranks the
+        // mismatched dictation locale's neutral score.
+        let winner = LanguagePicker.pick(from: [
+            LanguageCandidate(localeID: "en-US", meanConfidence: 0.85, transcriptLength: 20),
+            LanguageCandidate(localeID: "ru-RU",
+                              meanConfidence: LanguagePicker.neutralConfidence,
+                              transcriptLength: 22),
+        ])
+        #expect(winner?.localeID == "en-US")
+    }
+
+    @Test func twoNeutralDictationCandidatesBreakByLength() {
+        // Both engines dictation-only (e.g. ru-RU vs uk-UA), both neutral-scored:
+        // the longer transcript wins the tie-break.
+        let winner = LanguagePicker.pick(from: [
+            LanguageCandidate(localeID: "uk-UA",
+                              meanConfidence: LanguagePicker.neutralConfidence,
+                              transcriptLength: 8),
+            LanguageCandidate(localeID: "ru-RU",
+                              meanConfidence: LanguagePicker.neutralConfidence,
+                              transcriptLength: 24),
+        ])
+        #expect(winner?.localeID == "ru-RU")
+    }
+
     // MARK: - Argmax
 
     @Test func pickChoosesHighestConfidence() {