speed up CI / test.sh

[jyoung8607]

Trying some things in CI that are working for me locally:

• There are two very long MISRA mutation tests that hold up pytest, run them in a separate lefthook job
• Auto-count CPUs available for the main pytest run, speeds things up on powerful machines
• Use the worksteal scheduler to keep all parallel runners busy for the main pytest run

These changes cut time test.sh almost in half on my machine. We'll see how CI runners like it. Depending on the CI runner, going to auto CPU counts may actually reduce the number of parallel jobs, but worksteal scheduling should get back any potential parallelism lost. We shall see!