Note that ext4/006 is not in the "auto" or "quick" group, which means it is not commonly run.   One of the reasons for that is that being in the "dangerous_fuzzers" group, it may crash or wedge the kernel, especially if you are running an older kernel.   On the flip side, it only runs ten passes of the fuzzer, which given that it is doing random fuzzing, isn't enough to really comprehensively find all problems.   There's nothing wrong with this per se, but to be honest, e2fuzz isn't actually a tool that I personally use.   It's been much more effective to let Vegard Nossum run American Fuzzy Lop set up for kernel file systems[1], since it uses genetic algorithms to do a much more efficient job than just using random fuzzing.

[1] https://events.linuxfoundation.org/sites/events/files/slides/AFL%20filesystem%20fuzzing,%20Vault%202016_0.pdf

All of this being said, e2fuzz probably doesn't make sense to be installed by default by most distribution packaging systems, since its meant for developers' use.   What I will probably do for Debian is to create a new package and package it as a separate "e2fuzz.deb", since most people wouldn't want/need it.   I think modifying e2fsprogs ebuild to install it only on test images makes the most sense.

it is not easy today to have an ebuild install sep files for base & test images.  if the test that it's needed for isn't terribly useful, then sounds like it'd be better to just skip it.