Remove all calls to rand() from tests

Problem

• Maintainers and developers are confused about varying positive test assertion counts for patches passing tests.

Goal

• Ensure that assertion counts can be trusted.

Details

• Running the same test suite on the same machine can yield a difference of hundreds assertions.
• The difference is caused by various tests that use rand(1, 10) for their initial setup (which means that test assertions are exponentially increasing for each additional item).

Solution

• Remove all calls to rand() from all tests.