Run benchcomp unit and regression tests in CI

[karkhaz]

This will ensure that benchcomp will continue to correctly parse future changes to Kani's output.

Checklist

• Each commit message has a non-empty body, explaining why the change was made
• Methods or procedures are documented
• Regression or unit tests are included, or existing tests cover the modified code
• My PR is restricted to a single feature or bugfix

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 and MIT licenses.

[celinval]

Why do you need to build Kani for benchcomp-tests?

[karkhaz]

One of the regression tests (and probably more in the future) run Kani on two of the performance benchmarks. So it doesn't do the full perf test, but just runs benchmarks to ensure that it can continue to parse Kani's output in case the output changes later.

I could split the tests into regression tests that do and don't run Kani, but overall the Kani build time is going to dominate so it will take just as long anyway...

We may wish to add different regression tests for other kinds of parsers later. If we're looking to parse CBMC results, I have a patch that adds a litani parser and a regression test that runs litani (not on real proofs, again just something trivial that tests the parser)