Thorough tests for eval

[wwared]

As a follow-up to #139, we also want:

• Full error test coverage, one test for each source of errors from the Lair functions (i.e. testing all error-generating code-paths)
  • Relatedly, make sure that we have one test for every EvalErr that can be generated by Lurk evaluation (this is a subset of the above)
• Test functions that generate and verify full proofs for all tests. These can be #[ignore]d by default but should run in CI