Feature request: Support multi‑label filters when populating Grafana variables

[anilmoris]

impacted version: v6.5.2

Problem
When building Grafana variables with the OCI Metrics data source, the variable query helper dimensions(...) does not provide a way to scope the variables return values. This makes it impossible to scope a downstream variable to selections made in upstream variables.

for example, restricting $target to series where datacenter="$datacenter" and cluster="$cluster"

Why this matters
• Large compartments/tenancies produce very large, noisy variable lists when they can’t be constrained.
• Hierarchical dashboards (Datacenter → Cluster → Namespace → Pod/Hostname/Target) need variables that honor upstream selections.
• Lack of multi‑label scoping forces per‑panel query hacks and hurts maintainability/performance.

Current behavior
• dimensions($region, $compartment, $metric_namespace, , $resourcegroup) returns dimension tuples but has no way to pass a label selector such as {datacenter="$datacenter", cluster="$cluster"} to constrain what is returned.

Expected behavior
Support scoping variable queries by multiple labels.

examples:

dimensions($region, $compartment, $metric_namespace, , $resourcegroup,
selector='{datacenter="$datacenter", cluster="$cluster"}')

Workarounds tried
• Post‑filtering with regex in the variable, example:
Regex: /.datacenter="([^"])"./
Regex: /.target="([^"])"./

This regex filter only helps for one dimension and still fetches everything.

Proposed solution
• Add multi‑label filter support to dimensions(...) and/or dimension_values(...) using a Prometheus‑style {label="value", ...} selector.