fix: Pin metrics format so the mg-exporter keeps receiving JSON

[colinbarry]

Memgraph 3.13 changed the default of --metrics-format from JSON to OpenMetrics (memgraph memgraph#4678). Neither chart set the flag on the mg-exporter path, so both relied on that default; after upgrading, the exporter gets OpenMetrics it cannot parse and no metrics reach Prometheus, with --metrics-format=JSON per instance as the only workaround. Both charts now pin the format to match the scrape path they deploy, in both directions, so neither depends on the image's default.

Also adds scripts/verify-metrics-format.sh (8 helm template assertions, no cluster, wired into both lint workflows) and fixes the metrics check in verify-remote-monitoring.sh, which asserted up{job="memgraph-exporter"}: true even when the exporter parses nothing, which is why its existing mg-exporter leg passed against this bug.

[colinbarry]

Tracking

• [Link to Epic/Issue]

Standard development

• Update / add chart templates, values, and NOTES.txt as needed
• Run helm lint and helm template locally against the changed chart(s)
• Verify the chart installs and upgrades cleanly on a test cluster

Labels checklist

• Add a docs label (exactly one): docs-changelog-only, docs-needed, or docs-not-needed
• Add at least one component label: memgraph, memgraph-ha, memgraph-lab, or infrastructure
• Add at least one type label: bug, feature, or infrastructure
• Assign the PR to a milestone
  • If not known, set for a later milestone

Documentation checklist

• Write a release note, including added/changed clauses
  • Pinned --metrics-format=JSON when the mg-exporter is enabled so it keeps receiving JSON on Memgraph 3.13, which now defaults to OpenMetrics; upgrading rolls the pods and lets you drop any hand-added flag. #285
• [ Documentation PR link memgraph/documentation#XXXX ]
  • Is back linked to this development PR