fix: propagate signal exit code from the npm bin wrapper

[Volkov Daniil (codeGaus)]

What does this PR do?

backend/cli/bin/openscience (the npm launcher) ran the platform binary via spawnSync. When that binary is killed by a signal, result.status is null and result.signal holds the signal name — but the wrapper did typeof result.status === "number" ? result.status : 0, so it exited 0 and reported success for a killed process (e.g. an OOM kill or Ctrl-C). This exits with the conventional 128 + signal number (falling back to 1), so shells, callers and CI observe the failure.

How did you verify your code works?

node --check on the wrapper passes. The change follows the standard shell convention (128 + signum) using os.constants.signals, which is already required in the file. The launcher is a plain executable script with no existing unit coverage (its behavior is process-level), so no unit test was added.

Checklist

• bun run typecheck — n/a (plain JS launcher, not part of the TS build)
• bun test — n/a (no coverage for the launcher; signal handling is process-level)
• node --check passes
• Linked issue (none found)
• Screenshots (n/a)

[vercel]

Volkov Daniil (@codeGaus) is attempting to deploy a commit to the InkVell Team on Vercel.

A member of the Team first needs to authorize it.

[Aayam Bansal (aayambansal)]

Thanks for catching this. We re-landed the same fix as a first-party commit in #158 (exit 128+signal on a signal-kill instead of masking it as 0). Appreciate the report.