Made one commit that causes the test to fail normally as it is on clone-methods, rather than blowing everything up entirely. 2 failures, the same exact as on clone-methods. Want to do some memory checking, because there is a risk some leaks were introduced, but conversely a lot of inappropriate frees were removed.

Note, this works under the assumption that git_{type}_free always frees any instances of the datatypes that it has pointers to as well, which as far as @johnhaley81 and I can tell is always the case.

OK, so all the tests run locally for me except clone ssh (which segfaults), but that fails upstream as well. This is ready for a quick review and then it can merge in.

You guys want to merge this in after CI is done?

Yup, should be maybe 10-20 minutes