refactor: implement robust connection cleanup and fix task leakage

[rajanarahul93]

Summary

This PR resolves a resource leak where downstream background tasks were not correctly terminated upon connection loss. It ensures that if any task associated with a connection_id exits, a signal is sent to abort all other related tasks.

Key Changes

• Task Cleanup: Added kill_signal triggers to the exit paths of the Notifier, Updater, Writer, and Share Monitor tasks.
• Bug Fix: Changed return to break in receive_from_downstream.rs to prevent protocol errors from bypassing the cleanup logic.
• Stability: Refactored safe_lock calls and Arc management to satisfy Rust's Send and move requirements without introducing redundant logic.

Verification

• Automated Tests: cargo test passed (8 lib tests, 1 integration test).
• Build: Successfully compiled with cargo build.
• Manual: Verified binary integrity with dmnd-client --help.

fixes #19

[Fi3]

this need a test that reproduce the issue, it should fail before the patch and pass after.