docs: add benchmark chart + publish independent latency data

- Added benchmark-chart.png: dual-panel dark-theme chart
  Left: latency comparison (Direct→A3M forced→A3M auto)
  Right: cost savings projection (1K to 1M queries/mo)
- Chart embedded in both README.md and docs/BENCHMARK.md
- Everything measured with llm-gateway-bench v0.2.0 (third-party tool)
- Full methodology and reproduce instructions in docs/BENCHMARK.md

Author: Das-rebel

README.md

@@ -1085,6 +1085,8 @@ Expert queries (legal, medical, complex reasoning) are routed to **premium** —
 
 Measured with [llm-gateway-bench](https://github.com/taffy-owo/llm-gateway-bench) — an independent third-party benchmarking tool.
 
+![A3M Router Benchmark](docs/benchmark-chart.png)
+
 | Scenario | TTFT | vs Baseline | What You Get |
 |:---------|:----:|:-----------:|:-------------|
 | **Direct to Groq** (no gateway) | **138ms** | — | Raw provider speed |

docs/BENCHMARK.md

@@ -4,7 +4,9 @@
 
 **The answer:** +96ms for passthrough, +236ms for full intelligent routing — on a 138ms baseline.
 
-Everything below explains exactly where those milliseconds go, and why they're worth it.
+![A3M Router Benchmark](benchmark-chart.png)
+
+*Left: latency comparison. Right: cost savings projection. Dark theme.*
 
 ---