Droidspaces

- real-time terminal output streaming via libsu CallbackList
- implemented auto-scroll to the terminal

- feat: Implement a conditional Seccomp shield that applies only to systemd containers on legacy kernels to prevent grab_super deadlocks and improve compatibility - That means you can run Docker/LXC in non-systemd containers like Alpine Linux in legacy kernels as well.

- feat: remove PR_SET_NO_NEW_PRIVS from seccomp setup to fix sudo in Alpine
- Fix DNS race condition and simplify DNS setup
- feat: Implement force cleanup logic in cleanup_container_resources to prevent hangs with unkillable processes by skipping sync and using lazy/force unmounts
- feat: Increase container stop timeout and delay SIGTERM fallback during container shutdown, Send multiple graceful shutdown signals concurrently

- feat(backend): remount /data with suid for directory-based containers on Android
- style(backend): silence redundant ENOENT warnings for optional namespaces 
- refactor(cli): added a new function to print the startup banner and showing a recommended kernel version if the current kernel is too old
- style(cli): update startup banner and silence default DNS logs
- style(backend): silence redundant cgroup namespace warnings on legacy kernels
- fix(cli): restrict foreground mode to strictly interactive terminals

- app: synced droidspaces kernel configuration to the latest available config 

- revert Android optimizations only if unmount is not skipped and no containers are running
- Require rootfs or rootfs-img for restart command

- Change the SELinux context of rootfs.img to `u:object_r:vold_data_file:s0` after formatting to avoid mounting errors in the sparse image installer
- Update the boot script to auto-detect all `.img` files under /data/local/Droidspaces/Containers and set their SELinux context to `u:object_r:vold_data_file:s0`

- If the host is Android and rootfs.img mode is used, set the SELinux context of rootfs.img to `u:object_r:vold_data_file:s0` before mounting to prevent mounting issues

Added custom DNS servers support in the container configuration menu

terminal: upgrade terminal proxy from select to epoll for better performance
mount: harden unmount_rootfs_img with retries and settle time
android: refactor iptables configuration logic into a clean loop
core: harden PID validation and implement shared OS-release parsing
core: implement dynamic container list allocation and use shared OS-release parsing
network: implement multi-DNS support and update CLI options
fix: ensure host-side resource cleanup on container boot failure in rootfs.img mode

Changelog
- Improve rootfs mount and unmount resilience on older kernels like Kernel 4.14 by adding retry logic for image mounting and extra sync calls during unmounting

- Completely re-wrote the sparse image installer and hardened the image integrity verifying process before mounting the created rootfs.img

undefined
Changelog
- fix container restart bug: implement restart marker and early mount detection
- fix: sync and sleep 200ms after repairing the rootfs image AND before mounting the image

- Added "Mask service" option to all 4 stages (Running, Enabled, Disabled, Abnormal) in the Systemd Services menu
- Removed the redundant "Running, Enabled, Disabled..." color legend bar from the top of the Systemd screen

- Implement and integrate cgroup attachment to ensure processes entering containers are correctly placed in the host's cgroup hierarchy for D-Bus/logind functionality and refine container-specific cgroup path creation