DX-596-together-gpu-clusters: sync with mintlify-docs#1165

[zainhas]

Syncs with togethercomputer/mintlify-docs#1165 ("DX-785: docs: add storage performance test documentation").

Changed docs that triggered this sync:

• docs/health-checks.mdx

Skill changes:

• Add Storage Performance to the automatic acceptance testing list in references/cluster-management.md so the reference reflects that new clusters are validated for storage throughput/integrity alongside GPU and NCCL checks.
• Add a new Storage subsection under Available Health Check Tests describing the fio-based sequential read/write + checksummed write/read-back test.
• Document the new Skipped result state and the Shared volume PVC not created message that Kubernetes clusters get until a shared-volume PVC exists, and point at the existing Storage section for the fix.

Skipped intentionally:

• Pass/fail threshold table values (10 GiB/s read, 5 GiB/s write): the reference doesn't carry the docs' threshold table for other tests either, so keeping it out preserves scope.
• The docs' "Slow data loading -> run Storage Performance + NVBandwidth" tip: the skill reference doesn't have a symptom-to-test triage table.

Generated by the Sync Skills Cursor Automation. Please review before merging.