perf: add application benchmarks

[MartinP7r]

Summary

Adds Criterion benchmarks for core application operations:

• Tree flattening (Node::flatten) - benchmarks with varying tree sizes (depths 3-5, breadths 3-10)
• Filter nodes (App::filter_nodes) - benchmarks with empty, short, and long queries
• ANSI code stripping - single string and batch (500 lines) benchmarks

Changes

• Add src/lib.rs to expose modules for benchmarking
• Add benches/app_bench.rs with new benchmarks
• Make tui::app module public
• Make strip_ansi_codes function public

Test plan

• cargo clippy passes
• cargo test - all 27 tests pass
• cargo bench --bench app_bench runs successfully

Running benchmarks

# Run all app benchmarks
cargo bench --bench app_bench

# Run quick benchmarks
cargo bench --bench app_bench -- --quick

# Run specific benchmark
cargo bench --bench app_bench -- tree_flatten