CVS-177198 LLM Memory Usage tests

[porlows1]

🛠 Summary

CVS-177198 Update long tests for LLMs memory usage
Changes in tests related files required by https://github.com/intel-innersource/frameworks.ai.openvino.model-server.tests/pull/2684

🧪 Checklist

• Unit tests added.
• The documentation updated.
• Change follows security best practices.
  ``

[Copilot]

Pull request overview

This PR updates the functional-test resource monitoring utilities to better support LLM long-tests focused on memory usage, including additional memory breakdown metrics and Windows support.

Changes:

• Extend Docker resource monitoring to capture additional memory metrics (private/anon vs cache) and add an efficient “sample all metrics” path.
• Introduce a Windows resource monitor that collects process memory stats via PowerShell and writes them to artifacts.
• Update OVMS run context to attach the correct resource monitor depending on Docker vs Windows execution.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
tests/functional/object_model/resource_monitor.py	Adds extra Docker memory metrics, introduces sample_all(), and implements a new Windows resource monitor with CSV logging.
tests/functional/object_model/ovms_instance.py	Attaches Docker vs Windows resource monitors based on runtime environment and available OVMS PID.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.

Suppressed comments (2)

tests/functional/object_model/resource_monitor.py:267

• When writing CSV files on Windows, opening the file without newline="" can produce extra blank lines due to newline translation. Use newline="" when opening the file for csv.DictWriter.

        log_path = Path(artifacts_dir, f"windows_stats_pid_{self.ovms_pid}.log")
        with log_path.open("w") as csvfile:
            writer = csv.DictWriter(csvfile, fieldnames=self.FIELDS)

tests/functional/object_model/ovms_log_monitor.py:389

• get_all_logs() always calls self._proc.get_output() first, which drains stdout from the shared stream queue via pop(). That defeats the mirror-file logic below (and contradicts the comment) because other monitors / log-saving code will still miss those stdout lines even when a mirror file exists. Reorder the logic so stdout is only popped when there is no mirror, and in the mirror case only drain stderr (if needed) while reading stdout from the mirror tail.

    def get_all_logs(self):
        stdout, stderr = self._proc.get_output()
        if stderr: