fix(seo): define a bounded canonical URL strategy for NPM Stats

Bắt đầu với src/routes/stats/npm/index.tsx, đặc biệt là phần xác thực route và staticData.includeSearchInCanonical, sau đó lần theo hành vi của URL canonical, dữ liệu có cấu trúc, liên kết nội bộ và sitemap. Xác định mô hình URL có phạm vi giới hạn, phân biệt ngữ nghĩa với cách trình bày, trước khi triển khai mô hình đó, đồng thời thêm các bài kiểm thử cho các tổ hợp mặc định, package, scoped-package, so sánh, chỉ UI và không hợp lệ. Hoàn thành có nghĩa là chỉ những URL ổn định, có ý nghĩa mới có thể được lập chỉ mục và siêu dữ liệu của chúng phải khớp.

Problem

The clean /stats/npm URL currently redirects to a URL that serializes the complete default chart state. In a live audit, the resulting canonical URL was 2,355 characters long.

The route validates defaults for packageGroups, range, chart type, transforms, playback controls, legend state, dimensions, and other UI settings, then staticData.includeSearchInCanonical causes all of those values to become canonical. Each user interaction can therefore create another indexable URL.

Relevant route: src/routes/stats/npm/index.tsx

Google warns that parameter-driven faceted navigation can generate effectively infinite crawl spaces and slow discovery of useful pages: https://developers.google.com/crawling/docs/faceted-navigation

Goals

• Keep NPM Stats URLs shareable.
• Index only clean, stable pages with meaningful search intent.
• Prevent display and playback controls from creating indexable canonical variants.

Proposed direction

• Make /stats/npm a stable 200 URL with a clean self-canonical, or normalize to another intentionally short canonical URL.
• Omit default values from serialized search parameters.
• Separate semantic state from presentation state.
• Treat parameters such as range, transform, bin type, view mode, chart type, sorting, dimensions, playback, legend visibility, and similar controls as non-indexable UI state.
• Define a bounded format for package or package-comparison pages that should be indexable, potentially using stable path-based package identifiers rather than encoded JSON.
• Canonicalize or noindex non-semantic variants according to the chosen model.
• Ensure scoped package paths resolve correctly and have one canonical encoding.
• Make WebApplication JSON-LD url and any BreadcrumbList URLs match the selected canonical.
• Review FAQPage JSON-LD. Google does not normally show FAQ rich results for non-government and non-health sites, so retain it only if another consumer justifies the maintenance.

Acceptance criteria

• Loading /stats/npm does not redirect to a URL containing the full default chart configuration.
• The default canonical is short, stable, and self-referencing.
• Default values are omitted from generated links.
• Presentation-only variants do not self-canonicalize as independently indexable pages.
• Any intentionally indexable comparison page has unique SSR title, description, H1, canonical, and structured data based on the compared packages.
• Nonsensical or empty package combinations return an appropriate 404 or noindex response rather than another indexable state.
• Tests cover default state, one package, scoped packages, multi-package comparison, UI-only controls, and invalid combinations.
• Internal links and sitemap entries expose only the bounded set of indexable URLs.