Architecture: Implement proper POSIX signal handling (SIGINT / SIGTSTP / SIGCHLD)

[exlier]

Problem

• Signal handling is ad-hoc across the codebase. SIGINT/SIGTSTP are not consistently routed to the correct process group; SIGCHLD handling is incomplete, causing missed child status updates and occasional zombies.

Impact

• Ctrl-C/Ctrl-Z behave inconsistently. Job control (fg/bg) is unreliable. Child processes may become zombies or their exit/stop/continue events are missed, causing incorrect job states.

How to reproduce

• Run a foreground pipeline that spawns multiple processes and press Ctrl-C / Ctrl-Z repeatedly.
• Start background jobs and rapidly start/stop others; check for zombies with ps or wait failures.

Acceptance criteria

• Shell consistently manages process groups (setpgid/setsid and tcsetpgrp usage) so job control works like POSIX shells.
• SIGCHLD is handled reliably: all child exits/stops/continues are observed and the internal job table is updated; no zombie processes are left.
• No unsafe or non-async-safe functions are executed directly in signal handlers.
• Include tests or a test harness that simulate multiple child lifecycles and validate behavior.