[vernonstinebaker]

The #5 merge spliced its new host test onto #9's refactored test region in config.zig, leaving pre-0.16 tmp.dir.writeFile/realpathAlloc forms that broke zig build test on main (PR #5's CI check predated #9, so the merge result was never re-validated — the empty-suite gap documented in the CONTRIBUTING PR).

Repair uses the std_compat.fs.Dir.wrap form matching #9's other tests. 19/19 tests pass locally (18 from #9 + the host-default test from #5, now actually running). Fixes main's build.

[vernonstinebaker]

@DonPrus @sleep3r

Urgent support request.

In my eagerness to get the NullClaw projects flowing again, I've performed a merge that ended up breaking the NullTickets build. This PR fixes that mistake. Should be a quick/simple 3 minute review.

I'll be more careful going forward.

Thanks in advance for the support.

[DonPrus]

Reviewed against current main and the #5/#9 merge history. The fix is necessary: I reproduced main's Zig 0.16 compilation failure at the unwrapped tmp.dir.writeFile call. Both replacements reuse the existing compatibility layer and match the adjacent configuration tests; a repository-wide search found no remaining old-form temporary-directory writeFile/realpathAlloc calls outside the compatibility implementation. The PR contains no unrelated changes and merges cleanly.

Validation on f0dd14d: zig build passed, zig build test --summary all ran and passed all 19 tests, and bash tests/test_e2e.sh passed 96 checks. No blocking findings.