krb5: add a fix for broken configure test

[mistydemeo]

• Have you followed the guidelines for contributing?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?

---

This fixes krb5 on macOS 11, both ARM and x86_64. It turns out that clang's new pickiness about calling functions without their header definitions broke this test. The test is actually testing destructor attributes, and for convenience is using the C unlink function. That is defined in unistd.h, but the configure test doesn't #include this header.

I'll be submitting an upstream fix as well, but this inreplace fix gets us going.

[claui]

The CI failures only affect stellar-core, and are timeouts due to a long-running self test.

Both krb5 and stellar-core are security-sensitive formulae, and certainly deserve more attention than most other formulae in that respect. However, the fact that the 10.14 CI run passed is strong evidence that the timeout is not related to krb5. I think we can safely merge this.

[claui]

Thanks @mistydemeo! 🍻