Wrong content-length header for datasets

Describe the bug
When downloading the datasets from https://x.com/i/communitynotes/download-data using wget, it hangs, not receiving more data, because the content-length header is too big (566M) for the file being served (185M).

To Reproduce

wget https://ton.twimg.com/birdwatch-public-data/2024/09/07/notes/notes-00000.tsv

wget https://ton.twimg.com/birdwatch-public-data/2024/09/07/notes/notes-00000.tsv
--2024-09-07 13:43:04--  https://ton.twimg.com/birdwatch-public-data/2024/09/07/notes/notes-00000.tsv
Resolving ton.twimg.com (ton.twimg.com)... 2606:2800:233:7ee2:97c:ab4c:6c70:be36, 152.199.21.140
Connecting to ton.twimg.com (ton.twimg.com)|2606:2800:233:7ee2:97c:ab4c:6c70:be36|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 593120688 (566M) [text/tab-separated-values]
Saving to: ‘notes-00000.tsv’

notes-00000.tsv         32%[=======>                   ] 185.19M  --.-KB/s    eta 2m 58s

Expected behavior
The content-length header should be set to the file size.

Start by running the listed wget command against the dataset URL and compare the Content-Length header with the bytes actually served. The issue is done when the header matches the downloaded file size; the payload does not identify a repository file or entry point to change.