Add "run benchmark" command support

[bkontur]

I am just thinking, as a part of fellows repo issue, I would like to add CI pipeline for testing "benchmarks" - like we have short-benchmark* jobs in polkadot-sdk.
Encointer was merged to fellows without even noticing that benchmarks do not work see.
We've merged chain-spec-generator to fellows, so the only puzzle missing is how to run them without need to rebuild polkadot binary in fellows CI. E.g.

I think, there are two solutions:

1. release polkadot binary built with `--features runtime-benchmarks"
2. extend this try-runtime-cli - we are already using this binary for check-migrations, so maybe it could be possible to extend and just reuse this binary for running short benchmarks.

I think that testing/running benchmarks seems to be part of "try runtime" functionality :)

[bkontur]

@bkchr @ggwpez guys, wdyt?

[ggwpez]

so the only puzzle missing is how to run them without need to rebuild polkadot binary in fellows CI. E.g.

Yea i thought about this. It should be possible to build the benchmark pallet command as freestanding binary that works with any runtime.
But i dont see this as part of try-runtime. It can just be a freestanding bench runner binary.
For the other benchmarking commands its more difficult, since they actually use the node to author a block.

[liamaharon]

As @ggwpez mentioned, hopefully this will be added to the polkadot node soon.

Will try to keep try-runtime limited to operations related to calling into development runtime apis.