Problem

Teams evaluating NLS for production have no data on compile times, runtime performance, or known limits (e.g. max file size, recursion depth, number of @use modules).

Acceptance Criteria

• Benchmark suite measuring: parse time, compile time, runtime throughput for representative workloads
• docs/PERFORMANCE.md documenting results and known limits
• Benchmarks run in CI and tracked over time (regression detection)
• Documented: what happens when limits are exceeded (graceful error vs crash)

Why This Blocks v1.0

Engineering teams need to know what they're getting into before betting a project on it.