Don't re-store existing files

[makew0rld]

Currently the CID of the file is calculated in the webhook at the same time as the file is being written to disk, for efficiency. But this also means that if the file is overwritten on the disk, even if it's already been stored. Ideally this would not happen.

I'm not sure what the best fix for this is, as calculating the CID of the file in advance would require loading all of it into memory, and would be slow for large files.

This is a pretty minor issue for us for now.

[makew0rld]

This was achieved for Browsertrix in #67, since the Browsertrix webhook already gives us a hash that we can compare to the database.