Add BenchmarkDotNet benchmarks for representative render workloads

[davidnmbond]

Deferred from phase 8 step 8.3.6.

Scope:

• Add BenchmarkDotNet benchmarks for simple, medium, and complex documents
• Measure end-to-end render throughput and key sub-operations where useful
• Make the benchmark project runnable by contributors and CI-compatible if practical

Acceptance criteria:

• BenchmarkDotNet benchmarks exist in the repository
• Benchmarks cover at least simple, medium, and complex documents
• Results can be reproduced from documented commands