fix(runner): release the file teardown when Ctrl-C stops the run

[Chemaclass]

🤔 Background

Related #1323

Ctrl-C killed the children, swept the temp dirs and exited without running the interrupted file's tear_down_after_script. The per-test tear_down already survived, through the test subshell's EXIT trap, so the two halves of the pair behaved differently for one signal.

💡 Changes

• Settle the pending file teardown from the handler, reusing the debt --stop-on-failure skips tear_down_after_script in sequential runs #1321 introduced. After the kill, so the per-test hook still goes first as in a normal run; before the temp-file sweep, so a hook reading a bashunit::temp_file still finds it.
• Reset SIGINT to its default disposition first. The hook is user code and may never return, and a handler that cannot be interrupted would leave no way out but SIGKILL. Checked by hand with a hook sleeping 60s: a second Ctrl-C ends the run in about 2s.
• Covered by a unit test of the handler body. An acceptance test that signalled a real run failed 2 runs in 3 under a loaded --parallel suite, since delivery depends on job control and on which frame the shell is in when the signal lands.
• --parallel is untouched and still runs neither teardown on Ctrl-C. The parent cannot fix that, because several files are in flight and each file's hook is redefined as the loop advances. Filed as Ctrl-C runs neither teardown under --parallel #1331.