Run runtime benchmarks on dedicated machine

[HCastano]

At the moment benchmarks are run in a very ad-hoc way on developer laptops.

We should instead be running benchmarks on a dedicated CI machine which matches the reference hardware for Polkadot.

To start just having access to such a machine and manually running benchmarks there is enough. Ideally we get to a stage where we have a dedicated benchmarking bot/CI job as is done in the Polkadot-SDK repo.

When either of these gets updated the RELEASE_CHECKLIST should also be updated to reflect the new benchmarking workflow.

[HCastano]

We've moved towards spinning up a machine manually during releases.

This could be further improved with https://github.com/entropyxyz/devops-infrastructure/issues/17.