Support workload-defined metrics for function autoscaling

Description

Allow a function version to select a workload-defined metric and threshold for autoscaling. Metrics flow through the OpenTelemetry collector to a Prometheus-compatible store, where the NVCF function autoscaler evaluates them. Existing limits, cooldowns, stickiness, and scale-to-zero safeguards remain authoritative.

Cluster placement and request routing are outside this epic.

Definition of Done

• A user can configure a metric identity and threshold for a function version.
• Workload metrics reach the configured time-series store with tenant and function-version identity.
• The autoscaler uses the selected metric for scale-up and scale-down decisions.
• Missing, stale, invalid, or ambiguous data fails safely and cannot cause unsafe scale-down.
• Functions without this configuration retain existing autoscaling behavior.
• Decision telemetry identifies the metric source and decision reason.
• End-to-end tests and user documentation cover configuration, operation, and troubleshooting.

Resources (optional)

• Function autoscaler integration
• Function autoscaler documentation

By submitting this issue, you acknowledge that you are an assigned member of the NVCF development team and agree to follow our code of conduct and our contributing guidelines.