fix(types): align WebAudioContext.decodeAudioData signature with weap…

…p docs

Updated the `decodeAudioData` method in `WebAudioContext` to match the official WeChat Mini Program documentation:
- Modified the return type to `Promise<AudioBuffer>` instead of `AudioBuffer`, because of this method' actual response.
- Adjusted method parameters to accept `audioData`, `successCallback`, and `errorCallback`.

This fixes the TypeScript error and aligns the implementation with the expected API behavior.
Closes NervJS#17080

packages/taro/types/api/media/audio.d.ts

@@ -740,7 +740,11 @@ declare module '../../index' {
      * ```
      * @see https://developers.weixin.qq.com/miniprogram/dev/api/media/audio/WebAudioContext.decodeAudioData.html
      */
-    decodeAudioData(): AudioBuffer
+    decodeAudioData(
+      audioData: ArrayBuffer,
+      successCallback: (buffer: AudioBuffer) => void,
+      errorCallback: (error: any) => void
+    ): Promise<AudioBuffer>
   }
 
   namespace WebAudioContext {