Automated benchmark testing

[conradtchan]

The benchmarks are not currently tested until they are run in the main Phantom repo. This can cause test failures that are only caught very late in the testing process.

For example, #12 was merged before it was ready, causing this test failure for an unrelated commit on the main Phantom repo.

Checking if the benchmark passes with a workflow in this repo will reduce the likelihood of this issue.

[danieljprice]

this is a good idea