Refactor to reuse Podman client for HTTP requests in runtime

[youngtaekiim]

This pull request refactors the HTTP client usage in the Podman runtime module to improve efficiency and code clarity. The main change is that the code now uses a single, shared hyper::Client instance for all requests, instead of creating a new client on each call. This reduces resource usage and improves performance.

Podman HTTP client refactor:

• Introduced a static PODMAN_CLIENT using once_cell::sync::Lazy, so the hyper::Client is created only once and reused for all get, post, and delete requests.
• Centralized the Podman socket path in a single constant PODMAN_SOCKET, removing repeated code and making it easier to update the socket path if needed.
• Simplified the get, post, and delete functions to use the shared client and socket constant, reducing duplication and potential for errors.

[Chulhee1Lee]

OK.

[Copilot]

Pull request overview

Refactors the Podman runtime HTTP helper layer in nodeagent to reuse a single hyper::Client instance across requests, reducing per-call client construction overhead and centralizing the Podman socket path.

Changes:

• Introduced a shared, lazily-initialized PODMAN_CLIENT (once_cell::sync::Lazy) for all Podman HTTP requests.
• Centralized the Podman Unix socket path into a single PODMAN_SOCKET constant.
• Updated get/post/delete helpers to use the shared client and socket constant.