Improve fuzzing setup

The basic setup is there, but the fuzz corpus collection should be ported from https://github.com/FerretDB/FerretDB/blob/main/cmd/envtool/fuzz.go (probably to https://github.com/FerretDB/fuzz-corpus itself to avoid duplication of code with FerretDB?)

We probably should also make our fuzzing data smaller (see [there](https://afl-1.readthedocs.io/en/latest/tips.html#keep-your-test-cases-small)).