feat(amd): add device-wide PM sampling

[CodingInAVan]

Description

• Add an AMD device-counting profiling engine.
• Use GPUBusy as the portable default metric.
• Support explicit native metrics through pm_sampling_metrics.
• Respect PM sampling cadence, sample limits, scope gating, and deep-window
  gating.
• Attribute PM rows to scopes and the correct AMD device.
• Report device-counting capability status based on collected PM rows.
• Preserve baseline tracing when device counting is unavailable.
• Add the self-checking amd_pm_sampling_sample_rows example.
• Improve AMD report presentation:
  • recover anonymous-namespace kernel names
  • use Compute Units, LDS, Wave Occupancy, and Waves/CU
  • use backend-neutral scope timing language
  • hide metadata-only instruction analysis

Type of Change

• Bug fix
• New feature
• Documentation update

Testing

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas