e2e-tests: Always run tests with all compatible runners

[justinvreeland]

It doesn't take that much longer to run these tests with the additional runners. It would be nice to have a switch to select that but I didn't want to spend too much time on it. This is roughly on top of #2134 but has divered once I realized I wouldn't be able to fix the failing tests easily.

For example on x86_64 linux we have the following tests that fail with both the same for x86_64 testing and aarch64 testing/

FAIL:  env-vars-build-test.yaml/build/docker git-checkout-build.yaml/build/qemu git-checkout-build.yaml/build/docker xcover-nopkg-test.yaml/test/qemu xcover-nopkg-test.yaml/test/bubblewrap xcover-nopkg-test.yaml/test/docker

[justinvreeland]

Looks like CI has a different set of failures than I see locally though:

2025/08/20 15:13:43 WARN + exit 0
2025/08/20 15:13:43 INFO pod 9a17ba24ab17cd13e28fb25c4b2d8523b87eaec4704d160063bfcd04045835e1 terminated
FAIL:  env-vars-build-test.yaml/build/docker git-checkout-build.yaml/build/docker xcover-nopkg-test.yaml/test/bubblewrap
make: *** [Makefile:175: test-e2e] Error 1

[stevebeattie]

The difference you're seeing between CI and local may be due to running as non-root locally versus as root in the CI runner; both the bubblewrap and docker runners have subtle differences in behavior in those situations due to the outside world leaking in.

I do like this change, I wanted to run against all three runners when developing #2138 and also want to see them get run in CI.