[P1][SEO] Reconcile sitemap URLs, aliases, and Search Console indexing

[sbc1-code]

Baseline (authenticated Search Console, 2026-08-06)

• 118 indexed URLs.
• 186 not indexed: 117 redirects, 17 legacy 404s, 43 discovered/not indexed, 5 crawled/not indexed, 2 noindex, 2 robots-blocked.
• Sitemap status: Success; last read 2026-08-05; 158 pages discovered.
• Current local build emits 178 sitemap URLs.
• 71 HTTPS URLs, zero HTTPS issues.
• 70 valid breadcrumb items and 71 valid Dataset items, zero invalid.
• 71 mobile URLs are Good in Core Web Vitals; zero poor or needs-improvement URLs.

Several 404 examples now return 200 or canonical 301s, so Search Console needs evidence-based reconciliation rather than blanket URL creation.

Scope

• Crawl every URL in the generated sitemap and every alias/redirect inventory entry.
• Verify status, final URL, self-canonical, language/hreflang, indexability, and exactly one intended canonical.
• Classify each Search Console 404 as fixed alias, intentional removal, or real defect.
• Fix only real defects; do not revive spam/obsolete paths.
• Submit/validate the corrected groups in Search Console and record the baseline date.
• Investigate why the 43 newly discovered URLs have not been crawled, starting with the broken /best-time/ hub and internal-link reachability.

Definition of done

• Automated crawl report has zero accidental canonical 404s, loops, or conflicting canonicals.
• Sitemap count matches the canonical route inventory.
• Fixed 404 groups are submitted for validation.
• Discovered/crawled-not-indexed URLs have a documented keep/improve/remove decision.
• Recheck is scheduled after Google has had time to recrawl.

Parent roadmap: #56

[sbc1-code]

Root cause removed, shipped 2026-08-20 in #73. This issue can now be scoped down to the GSC reconciliation pass.

The reason sitemap URLs and indexed URLs kept diverging was not a sitemap generation bug. CBP intermittently substitutes a different row for the same physical crossing (different port_number AND port_name), and slugs were derived from names, so a crossing's URL moved on its own. Verified across 200 committed snapshots: the pairs are mutually exclusive, 0 co-occurrence.

Measured before the fix: 9 distinct slug maps across 200 snapshots, 12 unstable ports, /crossing/paso-del-norte/ serving a hard 404 while the sitemap advertised el-paso-paso-del-norte-pdn, and ~22% of deploys failing.

After: 1 distinct slug map across all 200 snapshots. Slugs are pinned per port in BASE_OVERRIDES, so baseSlug(name) is now only a fallback for genuinely new ports. Three previously-unpatched aliases were added and all resolve 200 in production.

Remaining work for this issue: a GSC pass to retire the stale indexed URLs and confirm the 42 canonical crossing pages are the ones indexed.