Investigate if we can reduce the number of tests we run on the macos bot

[kenzieschmoll]

The macos bot takes much longer to run than the bots that run on linux. Here's an example where the tests took ~12 min to run on linux and ~26 min on macos: linux: https://github.com/flutter/devtools/actions/runs/3206576642/jobs/5240498195 macos: https://github.com/flutter/devtools/actions/runs/3206576642/jobs/5240498273

We really only need the macos bot to run our golden image tests (and maybe some integration tests if we create proper integration tests at some point). We should investigate if we can shard up our tests so that we can reduce the run time of the macos job.