Fixup libafl_fuzzbench-based fuzzers

[addisoncrump]

Fixes #2019.

[addisoncrump]

This PR also restores the mruby bug benchmark. Potentially I could bring back other benchmarks for these fuzzers.

@DonggeLiu, would you think it appropriate to run a fuzzbench run for these fuzzers + a working symbolic fuzzer + baselines (AFL++ and libafl at least) against mruby? After #2028 lands.

[DonggeLiu]

Yep sure, I would love to see the results too. Thanks for doing this, @addisoncrump!

If you are interested in running them on more bug-based benchmarks, they (and their results) can be found on sbft23 branch and #2022. @Dammo3 kindly added some benchmarks recently.

Two minor things:

1. We conventionally append the commit ID after the directory name of bug-based benchmarks, e.g., https://github.com/google/fuzzbench/tree/master/benchmarks/bloaty_fuzz_target_52948c

2. How about changin the type to type: bug:

   fuzzbench/benchmarks/bloaty_fuzz_target_52948c/benchmark.yaml

   Line 20 in 782b361

   type: bug

[addisoncrump]

@DonggeLiu Maybe something worth noting: the CI will not fail if the benchmark doesn't exist!

[DonggeLiu]

Wow, I did not know this, thanks!
I reckon this is low priority because it does not affect checking fuzzers or running experiments.
WDYT?

[addisoncrump]

I reckon it's low priority as well, but maybe something to keep in mind since it might inadvertently mark a fuzzer as "working" for a non-existent benchmark.

[DonggeLiu]

Yep, created this: #2037

[renatahodovan]

@addisoncrump Thanks for working on this! Is it planned to merge it anytime soon?

[addisoncrump]

This is ready to be merged.