Don't try to check accept rate in standard particles test

The test was brittle against makers that apply selections, i.e. that go beyond the reconstruction, and has been failing for a while because of the addition of such makers.

We don't have a good way to evaluate the physics performance of individual makers yet, so rather than trying something ad-hoc just assert the makers can be run over data.

---

Also changes the CMTCONFIG value used in the CI test to track the GCC 8 → 9 migration.