As you can see it took only a second to download and emitted only one byte to stdout, but it downloaded and verified the integrity of the 128 KiB segment that contained that byte.
It takes 1 second for YOU to download 128 KiB (and for your counterpart(s) to upload 128 KiB). Also this generates heavy traffic for nodes - I would test them once per hour or even more often. This means there would be significant outgoing traffic if I host a few TB only for testing that I still have the files.
CPU time is cheaper than bandwidth. It would make more sense to precompute CRC32 oder Adler32 hashes of random ranges in each file, distribute these and request these hashes. Once you run out/low, you can do a full verify, download the file + generate new hashes. This could be tuned to happen once a month or even less frequent, depending on your personal rating on that node's reliability.