feat: parameterize environment variables in Helm chart

[ben-gutman]

Description

This PR parameterizes 15 hardcoded environment variables across the 3 DaemonSet templates (device-plugin, gpu-feature-discovery, mps-control-daemon), moving them to `values.yaml` for greater configurability.

Motivation

The chart currently hardcodes several ENV variables that users might need to customize for different Kubernetes environments:

• Config paths for alternative container runtimes
• Node labels for custom device plugin configurations
• NVIDIA device visibility and capability settings

This change makes the chart more flexible while maintaining full backward compatibility.

Changes

• values.yaml: Added `configManager` and `env` sections with 15 new configurable parameters
• templates/daemonset-*.yml: Updated init, sidecar, and main containers to use values from values.yaml
• Chart.yaml: Bumped version to 0.19.0

Backward Compatibility

✅ 100% backward compatible - All defaults match original hardcoded values exactly.

Users upgrading from 0.18.0 will see no behavior change unless they explicitly override values.

Testing

• `helm template` renders successfully with default values
• `helm template` renders successfully with custom overrides
• `helm lint` passes without warnings

Example Usage

  # values.yaml override example
  configManager:
    init:
      nodeLabel: "custom.io/device-config"
  env:
    nvidia:
      visibleDevices: "0,1"
      driverCapabilities: "compute"

Signed-off-by: Ben Gutman ben.gutman@gong.io

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.