Add fuzzing for the rust parser

[stusmall]

This adds a fuzzing task for the rust parser. This is run as release to avoid various debug assertions in the project. I've had this running on my workstation and it hasn't found anything, but it is useful to keep in CI to catch regressions.

I recongize that fuzzing adds extra time to the pipeline, so I did a few things to try and minimize the impact.

• First, it is a seperate, parallel workflow. Since the main build and test is a debug build and fuzz is a release build there is very little wasted work in rebuilding.
• Second, I added caches for both tasks. This should help keep build time town in general.
• The time the fuzzing task will run is capped at 2 minutes. This is pretty short. This was picked to not be too much longer than previous runs. This is a bit of a trade off. The longer this value is, the better the coverage.

Later I will take a look at having it excercise more parser configuration options to get better coverage. At first I just wanted to try and cover the default case well.

[dubadub]

Looks good to me, what do you think @Zheoni?

[Zheoni]

I'm not familiar with fuzz testing. I understand that it tries random inputs to check that the parser doesn't panic, is that correct?

If that's it I'm happy to merge this, caching the builds it's great too! 😄

[stusmall]

Yup. The idea is that it will generate random input and observes coverage. It mutates the input to try and maximize coverage. It's really useful for things like this that take broad, plaintext input and parses it. It helps shake out hidden panics, logic bugs, unicode mistakes, infinite loops and various crashes.

Fuzzers are generally fairly dumb and don't give you a lot of input on if the parsing was "correct", just that it didn't explode

[Zheoni]

Ok, seems like a good idea to have it. Thanks!