Improves signal handling

[arcanis]

We were only catching SIGINT but not SIGTERM. Additionally, while Yarn Switch was catching the signals, Yarn wasn't when proxying commands.

Fixes #236

---

Note

Medium Risk
Changes Unix signal handling for multiple command execution paths, which can affect how long-running processes terminate and what exit codes are propagated (especially under Ctrl-C or container shutdown). Scope is limited and covered by new acceptance tests, but regressions could impact CLI usability.

Overview
Improves signal delegation so wrapper/parent processes ignore both SIGINT and SIGTERM while waiting on spawned children, allowing the child process to handle shutdown and ensuring the child’s exit code is propagated.

This replaces IgnoreSigint with IgnoreSignals and enables enable_signal_delegation() across more execution entrypoints (e.g. dlx, exec, node, python, run), plus adjusts inherited-stdio script execution to wait on the spawned child consistently.

Adds acceptance tests for switch proxy to verify process-group SIGINT and SIGTERM are correctly delegated and result in clean exit code 0.

^{Reviewed by Cursor Bugbot for commit fbb67f4. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit bf6b5ad. Configure here.}

[github-actions]

⏱️ Benchmark Results

gatsby install-full-cold

Metric	Base	Head	Difference
Mean	4.464s	4.471s	+0.15% ⚠️
Median	4.458s	4.460s	+0.04% ⚠️
Min	4.290s	4.356s
Max	4.574s	4.626s
Std Dev	0.061s	0.067s

📊 Raw benchmark data (gatsby install-full-cold)

Base times: 4.491s, 4.495s, 4.507s, 4.393s, 4.449s, 4.440s, 4.512s, 4.435s, 4.412s, 4.457s, 4.486s, 4.485s, 4.492s, 4.454s, 4.431s, 4.342s, 4.456s, 4.390s, 4.416s, 4.457s, 4.574s, 4.538s, 4.550s, 4.523s, 4.509s, 4.495s, 4.533s, 4.290s, 4.459s, 4.450s

Head times: 4.533s, 4.377s, 4.452s, 4.429s, 4.409s, 4.462s, 4.483s, 4.413s, 4.458s, 4.447s, 4.504s, 4.417s, 4.418s, 4.356s, 4.453s, 4.393s, 4.386s, 4.571s, 4.486s, 4.558s, 4.437s, 4.538s, 4.397s, 4.571s, 4.543s, 4.487s, 4.485s, 4.549s, 4.626s, 4.486s

---

gatsby install-cache-only

Metric	Base	Head	Difference
Mean	1.269s	1.287s	+1.42% ⚠️
Median	1.266s	1.287s	+1.65% ⚠️
Min	1.227s	1.247s
Max	1.306s	1.323s
Std Dev	0.019s	0.019s

📊 Raw benchmark data (gatsby install-cache-only)

Base times: 1.292s, 1.303s, 1.267s, 1.227s, 1.262s, 1.261s, 1.268s, 1.281s, 1.277s, 1.288s, 1.250s, 1.275s, 1.267s, 1.291s, 1.284s, 1.306s, 1.305s, 1.255s, 1.256s, 1.278s, 1.257s, 1.250s, 1.263s, 1.252s, 1.269s, 1.259s, 1.265s, 1.240s, 1.251s, 1.265s

Head times: 1.290s, 1.272s, 1.282s, 1.262s, 1.263s, 1.288s, 1.304s, 1.302s, 1.311s, 1.298s, 1.302s, 1.323s, 1.286s, 1.291s, 1.309s, 1.295s, 1.316s, 1.303s, 1.309s, 1.284s, 1.288s, 1.273s, 1.285s, 1.261s, 1.271s, 1.270s, 1.256s, 1.247s, 1.283s, 1.282s

---

gatsby install-cache-and-lock (warm, with lockfile)

Metric	Base	Head	Difference
Mean	0.340s	0.351s	+3.10% ⚠️
Median	0.340s	0.349s	+2.61% ⚠️
Min	0.329s	0.338s
Max	0.355s	0.376s
Std Dev	0.007s	0.010s

📊 Raw benchmark data (gatsby install-cache-and-lock (warm, with lockfile))

Base times: 0.340s, 0.354s, 0.337s, 0.336s, 0.336s, 0.335s, 0.336s, 0.344s, 0.347s, 0.355s, 0.343s, 0.345s, 0.352s, 0.347s, 0.340s, 0.344s, 0.341s, 0.348s, 0.332s, 0.329s, 0.334s, 0.332s, 0.337s, 0.337s, 0.335s, 0.343s, 0.339s, 0.333s, 0.341s, 0.342s

Head times: 0.348s, 0.342s, 0.344s, 0.349s, 0.351s, 0.347s, 0.341s, 0.341s, 0.338s, 0.338s, 0.340s, 0.339s, 0.346s, 0.342s, 0.347s, 0.358s, 0.342s, 0.349s, 0.355s, 0.355s, 0.358s, 0.353s, 0.363s, 0.364s, 0.371s, 0.358s, 0.359s, 0.358s, 0.376s, 0.362s

[merceyz]

Tested this and can confirm it fixes the issue I was having.