feat: Add 'Build vs Buy in 2026' comprehensive guide (#980)

* feat: Add 'Build vs Buy in 2026' comprehensive guide

Adds third long-term blog post from Issue #868 covering build vs buy
decisions for infrastructure tooling.

Key sections:
- True cost formula (initial + maintenance + opportunity cost)
- 8 infrastructure categories analyzed (identity, CI/CD, observability,
  secrets, feature flags, load balancing, messaging, IDPs)
- Decision framework flowchart
- Real-world scenarios (Series A/B/C with actual cost calculations)
- Common mistakes and when building actually makes sense

All pricing verified for 2026, math calculations validated, includes
OG images for social sharing.

* fix: Correct cost calculations in Scenario A and C

- Scenario A: K8s setup cost corrected from $100K-$150K to $67K-$100K
- Scenario C: Initial build cost corrected from $600K-$1M to $900K-$1.6M
- Scenario C: 3-year TCO corrected from $3.2M-$4M to $3.3M-$4.6M
- Updated cost multiplier claim from 2-3× to 2-4× (accurate)

All calculations now align with stated engineer-months and $16.7K/month rate.

* fix: Correct SaaS pricing for Auth0, GitLab CI, and HashiCorp Vault

- Auth0: Changed from $240-$2,400/year (10K-100K MAU) to $500-$5K/year (10K-25K MAU)
  * Upper bound was 23× too low - 100K MAU actually costs ~$56K/year
  * Adjusted range to reflect more accurate small-medium scale pricing

- GitLab CI: Fixed from $19-$99/user/year to $19-$99/user/MONTH
  * Critical error - was off by 12×
  * GitLab Premium/Ultimate are priced monthly, not annually

- HashiCorp Vault: Changed from "$0.03/hour per secret" to "$0.30-$2.50/hour per cluster"
  * HCP Vault prices per cluster instance, not per secret
  * Annual estimate ($15K-$40K) remains accurate

All other pricing claims remain verified as accurate.

Author: bobbyonmagic