Restart fuzz-testing effort in rust

Hey team, I'm aware that you are likely busy with other work, is now a bad time to discuss the potential to restart the fuzz-testing effort?

__Use case:__
Fuzz-testing is very useful tool in discovering edge-cases and bugs that wouldn't otherwise be captured by the unit-testing suite.

__Proposal:__
1. Fuzz tests are rewritten using cargo-fuzz. I've got a proof of concept in #24501, to demonstrate what this would look like.
2. Port of the [google/oss-fuzz](https://github.com/google/oss-fuzz/blob/master/projects/influxdb/project.yaml#L2) configs to match the new rust build rather than the old golang build.

__Current behaviour:__
- No fuzz testing is currently being done (beyond ad-hoc RNG based fuzzing).
- The google/oss-fuzz influxdb [builds are broken](https://oss-fuzz-build-logs.storage.googleapis.com/index.html#influxdb) and have been since 2020.

__Desired behaviour:__
Continuous fuzzing fully re-enabled using google/oss-fuzz.

__Alternatives considered:__
- Continue using the [ad-hoc RNG based fuzzer's](https://github.com/influxdata/influxdb/blob/bb6a5c0bf6968117251617cda99cb39a5274b6dd/arrow_util/src/bitset.rs#L362-L403). These fuzzer's can often be hard to read, and typically operate in open-loop limiting the acheivable performance compared to a coverage-driven closed-loop fuzzer like libfuzzer. The nice thing is that the original RNG fuzzers can fairly easily be ported over to use libfuzzer instead, so they certainly weren't wasted effort.
- Continue using the [proptest style fuzzers](https://github.com/influxdata/influxdb/blob/main/cache_system/src/addressable_heap.rs#L557-L568), while the ergonomics/readability of these fuzzer's is significantly better they don't use coverage driven feedback, which limits the overall fuzzing performance.

Start with proof of concept PR #24501, then inspect the existing fuzzers in arrow_util/src/bitset.rs and cache_system/src/addressable_heap.rs. Review the referenced oss-fuzz project configuration and determine the required Rust cargo-fuzz and continuous-build changes. Done means coverage-driven fuzzing is restored and the google/oss-fuzz builds succeed.

Written by the indexing model from the issue text.