Stabilize test_linux_network_stacksmash_64 under Python 3.12 CI timing

[Copilot]

test_linux_network_stacksmash_64 has become materially flakier in CI after the move to Python 3.12. The failures are timing-driven: the generated exploit reaches the target path, but replay intermittently times out while the traced TCP service is still settling.

• Tighten startup timing around the traced target

  • Add a small delay when constructing the TCP rex.Crash so tracing does not begin before the service is ready.

• Give replay a stable startup window

  • Increase the wait before invoking the generated exploit script against the replay target.

• Reduce false negatives in exploit replay

  • Increase the subprocess timeout for the generated exploit so slower CI workers do not fail a valid exploit run during shell setup / command execution.

crash = rex.Crash(
    target,
    crash=inp,
    use_rop=False,
    aslr=False,
    input_type=rex.enums.CrashInputType.TCP,
    port=port,
    delay=1,
)

time.sleep(1)

exploit_result = subprocess.check_output(
    ["python", exploit_location, "127.0.0.1", str(new_port)],
    timeout=10,
)

[angr-bot]

Corpus decompilation diffs can be found at angr/dec-snapshots@master...angr/rex_121