CE: JSON-LD structured data for SERP rich results + AI crawlers

[ericflo]

What

Add JSON-LD structured data so search engines (Google rich snippets, Knowledge Graph) and AI crawlers (ChatGPT browsing, Perplexity) can properly understand ModelRelay.

• crates/modelrelay-web/src/templates.rs — page_shell_custom now renders an Organization JSON-LD script on every page it shells (dashboard, legal, integrate, setup, etc).
• crates/modelrelay-cloud/templates/index.html — landing page (served as static HTML, not via page_shell) gets both Organization and SoftwareApplication schemas.
• crates/modelrelay-cloud/templates/pricing.html — pricing page (also static) gets Organization schema.

Why

Verified gap before this PR: zero application/ld+json blocks on live site (curl -s https://modelrelay.io/ | grep -c ld+json returned 0 for both / and /pricing). JSON-LD enables SERP rich snippets, Knowledge Graph association, and AI crawler metadata extraction.

Note on implementation: the landing and pricing pages are served as static HTML strings (LANDING_HTML / PRICING_HTML), not through page_shell. So the Organization schema had to be added directly to those two static templates to hit the live-site validation targets. page_shell-shelled pages (dashboard, legal, setup, integrate, etc.) pick it up automatically through the page_shell_custom change.

Validation

• Added templates::tests::page_shell_includes_organization_jsonld which renders page_shell and asserts the Organization JSON-LD is present.
• cargo test -p modelrelay-web --lib → 1 passed.
• cargo test -p modelrelay-cloud --lib → 38 passed.
• cargo fmt --check → clean.
• cargo clippy -p modelrelay-web -p modelrelay-cloud --all-targets -- -D warnings → clean.
• Python json.loads confirms all rendered JSON-LD blocks parse as valid JSON (no format! brace-escape bugs).

Post-deploy:

• curl -s https://modelrelay.io/ | grep -c application/ld+json → expect 2 (Organization + SoftwareApplication)
• curl -s https://modelrelay.io/pricing | grep -c application/ld+json → expect 1 (Organization)