Add BenchmarkDotNet coverage for cloud architecture patterns

[JerrettDavis]

Goal\nAdd fluent-vs-generated BenchmarkDotNet coverage for cloud architecture patterns.\n\n## Scope\nInclude patterns such as Ambassador, Backends for Frontends, External Configuration Store, Gateway Aggregation, Gateway Routing, Leader Election, Scheduler Agent Supervisor, Sidecar, Strangler Fig, and any other cloud pattern in the catalog.\n\n## Acceptance criteria\n- Each benchmark has fluent and generated routes where supported.\n- Construction and execution phases are split when both costs matter.\n- Allocation and GC metrics are reported through the shared benchmark config.\n- Results are categorized by Cloud, pattern name, route, and phase.

[JerrettDavis]

Matrix baseline landed in #338 / 86e5a59. The full published result matrix now lists every pattern route and every generator source in docs/guides/benchmark-results.md, with TinyBDD checks preventing docs drift. Leaving this issue open for the deeper production-scenario timing work requested here: family-specific scenario benchmarks with realistic payloads, direct fluent-vs-generated timing, and allocation/GC comparisons beyond the route-coverage matrix.

[JerrettDavis]

PR #339 landed on main as 1f0977f with the next cloud benchmark slice.

Added dedicated BenchmarkDotNet fluent-vs-generated scenario classes and published current-TFM results for:

• Ambassador: construction 55.42 ns / 448 B fluent vs 48.03 ns / 360 B generated; execution 87.92 ns / 624 B fluent vs 93.72 ns / 624 B generated.
• Cache-Aside: construction 19.91 ns / 200 B fluent vs 19.85 ns / 200 B generated; execution 216.50 ns / 1,048 B fluent vs 208.60 ns / 1,048 B generated.
• Retry: construction 25.36 ns / 208 B fluent vs 27.18 ns / 208 B generated; execution 110.53 ns / 600 B fluent vs 109.52 ns / 600 B generated.

Docs updated in README, docs/guides/benchmarks.md, and docs/guides/benchmark-results.md. Added a TinyBDD drift check so every dedicated scenario benchmark class must have published construction and execution timing rows. PR and main checks are green, including CodeQL, CI, Publish docs, and Codecov.