Add accuracy off of test data.

[nkreeger]

Here is eval off of test data for class percentage. I also have one tweak to only return a pitch card for each pitch type. We can drop this if we don't like the cards.

---

This change is 

[dsmilkov]

---

Reviewed 4 of 6 files at r1.
Review status: 3 of 6 files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable