Makefile: build ginkgo before running localunit

[Animesh-Parashar]

Checklist

Ensure you have completed the following checklist for your pull request to be reviewed:

• I have read and understood our contributing guidelines and will not have more than two open PRs as a new contributor.
• PR description, commit message, and GitHub comments are human-written, per LLM Policy
• Certify you wrote the patch or otherwise have the right to pass it on as an open-source patch by signing all
  commits. (git commit -s). (If needed, use git commit -s --amend). The author email must match
  the sign-off email address. See CONTRIBUTING.md
  for more information.
• Referenced issues using Fixes: #00000 in commit message (if applicable)
• Tests have been added/updated (or no tests are needed)
• Documentation has been updated (or no documentation changes are needed)
• All commits pass make validatepr (format/lint checks)
• Release note entered in the section below (or None if no user-facing changes)

Does this PR introduce a user-facing change?

 None

On a fresh clone make localunit fails on clean checkout

/usr/bin/bash: line 1: ./bin/ginkgo: No such file or directory
make: *** [Makefile:675: localunit] Error 127

The current recipe runs $(GINKGO) which is ./bin/ginkgo but the target only declares test/goecho/goecho and test/version/version as prereq. Nothing builds ginkgo, so on a fresh clone the binary is missing. /bin/ is gitignored, so it is never checked in.

The other three targets that use ginkgo all declare .install.ginkgo: ginkgo-run (line 695), testbindings (715) and system.test-binary (766). localunit is the only one that does not.

But
hack/ci/runner.sh already works around this:

function run_unit() {
    make .install.ginkgo
    $SUDO make localunit
}

This makes that first line unnecessary.
I used a file target rather than adding .install.ginkgo to the prerequisite list, because .install.ginkgo is .PHONY and would rebuild on every invocation. That matters here: run_unit runs localunit under sudo, so an unconditional rebuild would produce a root-owned bin/ginkgo. With a file target, make skips the build when the binary is present, which I checked with
make -n on a warm tree (no rebuild commands emitted).

With the fix, ginkgo builds and 64 suites run. 63 pass. The one failure is pkg/machine/provider TestBadInstalledProviders, which needs qemu-system-x86_64 that I do not have installed locally, and is unrelated to this change.
make validatepr passes.

[TomSweeneyRedHat]

LGTM
assuming happy tests