Add support for testing both warnings and errors

[rstoneback]

Describe the requested feature
Current pysat tests generally test for warnings or errors, not both. Sometimes code produces both.

Pull #1179 adds some code that tests both warnings and errors. It also expanded the functionality of warnings tests. The 'no data at all' test in that pull could be the start of a function that does both.

This new combo function could be employed in test instruments, line 187, which states that current functionality can't test for both errors and warnings. A todo will likely be added.