[cosh-ng] feat: add a unified health check command

[kongche-jbw]

Component

cosh-ng

Problem Statement

cosh-ng already runs a startup health scan, and the Linux live collector covers host, CPU, memory, disk, kernel, and service state. Users cannot run a unified health check on demand, and provider, configuration, hook, PTY, and permission checks are not presented through one consistent interface.

Proposed Solution

Add an on-demand health check backed by a shared diagnostics engine:

• provide a cosh-shell doctor command;
• provide /health in the interactive shell;
• reuse the same checks and status model in both entry points;
• include host resources, kernel, services, provider readiness, configuration, hooks, PTY support, and permissions;
• classify results consistently as healthy, warning, or error;
• return stable exit status semantics;
• attach short, actionable remediation guidance to failed checks.

Alternatives Considered

Keep health checks limited to startup. Startup-only results are difficult to repeat, compare, or use in automated troubleshooting after the environment changes.

Acceptance Criteria

• Users can run health checks from both the CLI and interactive shell.
• Both entry points report the same checks and status semantics.
• Provider, configuration, hooks, PTY, and permissions are included.
• One failed check does not prevent the remaining checks from running.
• Statuses, exit codes, and remediation messages are stable and actionable.
• Tests cover healthy, degraded, partially unavailable, and failed environments.

[github-actions]

🔀 Issue Triage

This issue has been automatically assigned to @samchu-zsl as the maintainer(s) of cosh-ng.

Maintainers, please review and triage this issue. Set a priority label and update the status as needed. Thanks! 🙏

[BryanHeBY]

Hi, I would like to work on this issue. Could you assign it to me?