Improve coverage of parmest tests

[blnicho]

Documenting #1484 as an issue instead of a WIP PR.

Summary/Motivation:

• See discussion in Update data in Parmest example to match paper results #1474 about need for more robust tests

Changes proposed:

• Make test for using k_aug conditional on if the solver is install
• Update Rooney-Biegler test to use the correct data
• Test the examples, not just copy them to the test suit
• Verify k_aug and Pynumero report the same reduced Hessian
• Verify test for k_aug is skipped if it is not available
• Improve graphics tests (Improve ParmEst Graphics Tests #1346)