Add What Is AI Infrastructure? pillar page (#20881)

* Add What Is AI Infrastructure? pillar page

New /what-is/ pillar closing the top GEO gap on the AI Infrastructure
topic: a vendor-neutral category definition covering the accelerator,
data, orchestration, and serving layers competitors already cover, plus
the control-plane layer (infrastructure as code, agent-readiness,
governance) that they omit.

- Registers the page in data/what_is_sections.yml under Core concepts.
- Adds a cross-link from the existing ai-infrastructure-tools blog post
  to the new pillar to avoid two competing on-site definitions.
- Byline: Alex Leventer.
- Sourced from arXiv 2506.12270, platformengineering.org's 2026
  predictions, the 2025 DORA report, the CNCF 2025 Annual Cloud Native
  Survey, and a published Joe Duffy quote/stat from the-agentic-
  infrastructure-era blog post; all verified verbatim before use.

* Address pre-merge review findings on AI infrastructure pillar page

- Fix contradiction: HCL is a configuration language, not a general-purpose
  programming language; remove it from the 'real programming language' list
- Fix subject-verb agreement: 'patterns that show up' (was 'shows up')
- Style-guide substitution: use 'Pulumi Policies' instead of 'CrossGuard'
- Correct FAQ 30x-steps comparison to cite its actual baseline (CLI, not IaC)
  per the source paper's own table
- Ground the Kubernetes-for-AI-workloads claim in the CNCF figures already
  cited on the page instead of a blog link that doesn't make that comparison
- De-duplicate near-verbatim CNCF stat restatement between body and FAQ
- Fix blog cross-link framing drift: control plane governs the AI stack's
  own layers, not both of the post's tool categories
- Minor style-linter polish (wordiness, weasel word)

Verified against primary sources during review: the arXiv paper's own
step-count table (2506.12270), the CNCF/PR Newswire release, Google Cloud's
2025 DORA report, and Luca Galante's platform-engineering predictions post
all corroborate the page's other quoted figures verbatim.

* Fix possessive redundancy: 'Pulumi's Pulumi Policies' -> 'Pulumi Policies, Pulumi's implementation'

---------

Co-authored-by: workprentice <257153108+workprentice@users.noreply.github.com>

Author: workprentice[bot]

content/blog/ai-infrastructure-tools/index.md

@@ -62,7 +62,7 @@ The term covers two distinct categories that share almost no vendors.
 
 **AI-powered infrastructure management** is the inverse: AI tools that manage cloud infrastructure. They generate [infrastructure as code](/what-is/what-is-infrastructure-as-code/), run deployments, detect drift, and remediate policy violations. The pitch is that modern infrastructure (multi-cloud, containers, microservices, regulated workloads) has gotten too complex for humans to manage by hand and too varied for scripted automation to keep up with.
 
-Most organizations end up needing both: somewhere to run their ML workloads, and something to keep the rest of the cloud sane.
+Most organizations end up needing both: somewhere to run their ML workloads, and something to keep the rest of the cloud sane. For the full definition, including the control-plane layer that provisions and governs the rest of the stack, see [what is AI infrastructure](/what-is/what-is-ai-infrastructure/).
 
 ## Part 1: Tools for building AI infrastructure