docs: add unified Kubernetes monitoring entry point under install-pmm

[michaelcoburn]

PMM-XXXXX (JIRA ticket to be filed by Percona-Lab/pmm-workshop-pl2026 maintainer; this PR is the deliverable a JIRA ticket will reference)

Link to the Feature Build: SUBMODULES-0

Why

PMM Server can act as the centralized observability backend for both your databases and the Kubernetes clusters they run on — the integration with the Victoria Metrics Kubernetes monitoring stack already works and is documented in all four Percona operator docs repos:

• docs.percona.com/percona-operator-for-mysql/ps/monitor-kubernetes.html
• docs.percona.com/percona-operator-for-mysql/pxc/monitor-kubernetes.html
• docs.percona.com/percona-operator-for-mongodb/monitor-kubernetes.html
• docs.percona.com/percona-operator-for-postgresql/2.0/monitor-kubernetes.html

But PMM's own docs only mention this feature in passing on a dashboard reference page (reference/dashboards/kubernetes_monitor_operators.md), which is unlikely to be reached by a user searching for "how do I monitor Kubernetes with PMM."

This is a discoverability gap. A user reading PMM docs to plan their monitoring strategy has no entry point that says "yes, PMM can be your K8s observability backend too — here's how to plug it in."

What

Adds a single new page under install-pmm/install-pmm-client/connect-database/ that:

• Frames the use case (correlating DB-level and cluster-level signals in one PMM).
• Lists what gets captured by the stack (cAdvisor / kubelet / CoreDNS / kube-state-metrics / node-exporter / apiserver / control plane) in a table.
• Cross-links to the operator-specific monitor-kubernetes pages for the actual Helm walkthrough, rather than duplicating ~250 lines of procedure that would drift across five separate locations.
• Cross-links to existing PMM pages: service-account token, VictoriaMetrics reference, the Kubernetes overview dashboard, Helm-based PMM Server install.
• Marks the feature Technical Preview, matching the framing already on the dashboard-reference page.

Also:

• Adds the new page to the "connect-database" overview list (connect-database/index.md).
• Adds a "Kubernetes monitoring" subsection to the "Configure monitoring" nav in mkdocs-base.yml, alongside the existing Database / Cloud / System / Proxy / External-services subsections.

What this PR does NOT do

• It does not modify or remove the existing reference/dashboards/kubernetes_monitor_operators.md page — that page is left in place as the dashboard-reference entry point.
• It does not duplicate the operator-side Helm walkthroughs. Future improvements (e.g. consolidating to a single canonical procedure that the operator repos cross-link to, rather than each having their own copy) are out of scope for this PR.
• No code changes, no API changes — pure docs.

Test plan

• mkdocs build -f documentation/mkdocs.yml passes (couldn't validate locally — no mkdocs installed; relying on CI)
• All inter-doc relative links resolve to existing files (verified manually: api/authentication.md, reference/glossary.md, reference/third-party/victoria.md, reference/dashboards/kubernetes_monitor_operators.md, install-pmm-server/deployment-options/helm/index.md)
• Nav renders correctly: "Install PMM → Install PMM Client → Configure monitoring → Kubernetes monitoring → Connect Kubernetes cluster"
• Reviewer confirms the cross-links to operator docs are the most current canonical URLs (paths use docs.percona.com/percona-operator-for-*/monitor-kubernetes.html)

Context

This PR was generated as the deliverable for a docs audit performed for the upcoming Percona Live 2026 PMM 3 hands-on tutorial workshop. The audit confirmed that 7 out of 8 originally-suspected PMM 3 doc gaps had already been resolved by the docs team's PMM 3 refactor — this is the remaining one.

• API Docs updated (no API surface affected — docs-only)