feat: migrate on-device Qwen3 ASR to CoreML for background Flow dictation

Replace MLX GPU inference with CoreML bundles so transcription continues
while the host app is backgrounded. Adds model download and warm-up,
vendored Qwen3Speech, and updates onboarding, settings, and copy for the
~1.6 GB CoreML package (iOS 18+).

OSGKeyboardTests/UtteranceTranscriptStitcherTests.swift

@@ -0,0 +1,23 @@
+// UtteranceTranscriptStitcherTests.swift
+// OSGKeyboardTests
+
+import XCTest
+@testable import OSGKeyboardShared
+
+final class UtteranceTranscriptStitcherTests: XCTestCase {
+
+    func testMergeWithOverlapRemovesDuplicatedSuffixPrefix() {
+        let merged = UtteranceTranscriptStitcher.mergeWithOverlap(
+            previous: "今天天气很好",
+            next: "很好我们继续"
+        )
+        XCTAssertEqual(merged, "今天天气很好我们继续")
+    }
+
+    func testStitcherOrdersChunksByIndex() {
+        var stitcher = UtteranceTranscriptStitcher()
+        stitcher.append(index: 1, text: "第二段")
+        stitcher.append(index: 0, text: "第一段")
+        XCTAssertEqual(stitcher.composed(), "第一段 第二段")
+    }
+}