For type: pass-fail

• Apply {input, output}_validator_flags from testdata.yaml to the respective testcase when validating
• Collect testcase verdicts, infer testdata verdicts for each testgroup
• Update UI to report testgroup verdicts

For type: scoring:

• Allow testcase verdicts to have numerical score
• Apply grading flags from testdata.yaml to respective testcase scores
• Apply aggregation rules from grading in testdata.yaml
• Update UI to report aggregate scores for testgroups

Expectations:

• Read and parse expectations.yaml
• Compare testcase verdicts with expectations
• Compare testgroup verdicts with expectations