feat(core): add opt-in kill-on-drop for externally spawned commands

[hartsock]

Found during practical testing. We were experimenting with spawning shells and the outer process would terminate and leave behind the child shell.

sys::process::spawn never sets kill_on_drop, so a spawned child outlives the shell that started it. That is correct for a real shell — it is what makes job control, disowned jobs, and nohup-style usage work — and remains the default.

It is wrong for an embedded shell, where the host creates and destroys the shell as an object. There a surviving child keeps running unattended and holds a duplicate of the shell's stdout/stderr pipe, so a host draining that output never observes EOF.

Adds a kill_external_commands_on_drop creation option, defaulting to false, plumbed through RuntimeOptions to sys::process::spawn alongside the existing external_cmd_leads_session. The stub backend accepts and ignores it. Process-group behavior is untouched, so this signals the immediate child only.

Tests cover both directions and that the option defaults off. Mutation-checked: forcing kill_on_drop(false) fails the enabled case.

[hartsock]

Found alongside #1184 and #1242 during practical testing of brush embedded as a shell library.

[github-actions]

Public API changes for crate: brush-core

Added items

+pub brush_core::options::RuntimeOptions::kill_external_commands_on_drop: bool
+pub brush_core::CreateOptions::kill_external_commands_on_drop: bool
+pub type brush_core::ShellBuilderState::KillExternalCommandsOnDrop

Performance Benchmark Report

Benchmark name	Baseline (μs)	Test/PR (μs)	Delta (μs)	Delta %
clone_shell_object	17.57 μs	17.47 μs	-0.10 μs	⚪ Unchanged
eval_arithmetic	0.15 μs	0.15 μs	0.00 μs	⚪ Unchanged
expand_one_string	1.68 μs	1.68 μs	-0.01 μs	⚪ Unchanged
for_loop	32.05 μs	31.74 μs	-0.32 μs	⚪ Unchanged
full_peg_complex	57.01 μs	57.18 μs	0.17 μs	⚪ Unchanged
full_peg_for_loop	6.24 μs	6.22 μs	-0.02 μs	⚪ Unchanged
full_peg_nested_expansions	15.99 μs	16.12 μs	0.13 μs	⚪ Unchanged
full_peg_pipeline	4.29 μs	4.25 μs	-0.04 μs	⚪ Unchanged
full_peg_simple	1.81 μs	1.80 μs	-0.01 μs	⚪ Unchanged
function_call	3.70 μs	3.53 μs	-0.17 μs	⚪ Unchanged
instantiate_shell	54.77 μs	55.84 μs	1.08 μs	🟠 +1.97%
instantiate_shell_with_init_scripts	26110.64 μs	26038.27 μs	-72.38 μs	⚪ Unchanged
parse_peg_bash_completion	2100.55 μs	2116.29 μs	15.73 μs	🟠 +0.75%
parse_peg_complex	20.98 μs	20.71 μs	-0.27 μs	⚪ Unchanged
parse_peg_for_loop	2.04 μs	2.02 μs	-0.02 μs	🟢 -1.23%
parse_peg_pipeline	2.16 μs	2.08 μs	-0.08 μs	🟢 -3.56%
parse_peg_simple	1.09 μs	1.08 μs	-0.01 μs	⚪ Unchanged
run_echo_builtin_command	16.81 μs	16.63 μs	-0.18 μs	⚪ Unchanged
tokenize_sample_script	3.33 μs	3.43 μs	0.09 μs	⚪ Unchanged

Code Coverage Report: Only Changed Files listed

Package	Base Coverage	New Coverage	Difference
brush-core/src/options.rs	🟢 92%	🟢 92.06%	🟢 0.06%
brush-core/src/shell.rs	🟢 93.99%	🟢 94.85%	🟢 0.86%
Overall Coverage	🟢 75.77%	🟢 75.78%	🟢 0.01%

Minimum allowed coverage is 70%, this run produced 75.78%
Maximum allowed coverage difference is -5%, this run produced 0.01%

Test Summary: bash-completion test suite

Outcome	Count	Percentage
✅ Pass	1581	74.96
❗️ Error	17	0.81
❌ Fail	157	7.44
⏩ Skip	339	16.07
❎ Expected Fail	13	0.62
✔️ Unexpected Pass	2	0.09
📊 Total	2109	100.00

[reubeno]

Thanks for the contribution! Agreed that this is valuable for embedded shell hosts.