Critical: Race condition in /getTranscript causes potential data leaks

**Problem**
The `/getTranscript` endpoint currently uses `glob.glob` and `max(key=os.path.getctime)` to identify the downloaded subtitle file.

**The Logic Flaw**
This assumes that the file currently being processed is *always* the newest file in the `subtitles/` directory.
In a concurrent environment (multiple users), if User B's download finishes slightly after User A's download, User A's request thread will incorrectly identify User B's file as the "latest" and return the wrong transcript.

**Impact**
1. **Data Leak:** Users receive transcripts for videos they did not request.
2. **Crash:** If the "latest" file is deleted by another thread before it can be read, the server throws a `FileNotFoundError`.

**Proposed Solution**
Since the filename is deterministic (`subtitles/{video_id}.vtt`), we should explicitly construct the file path using the `video_id` instead of searching the directory.
I am working on a fix for this and will submit a PR shortly.