Faster "Benchmark" workflow with sharded CodSpeed uploads

[fargito]

Hello,

I'm a member of CodSpeed, and I saw that you were using it to run benchmarks.

We recently released a feature to enable sharded benchmark reports. Since on this repo, the "Benchmark" workflow can take between 45 minutes and 6 hours, I figured you could greatly benefit from this feature.

Basically, each crate can now run its own benchmarks independently. The whole aggregation will be done on CodSpeed infrastructure. Here, it would also mean that the benchmarks no longer have to run on the self-hosted instance. You can view the sharded benchmarks documentation here.

If you want, I can open a pull request to use this new approach in the "Benchmark" workflow. Also since I developed this feature, I'd be happy to help if there are any issues with it.