feat(wasm): add comprehensive WASM bindings for all algorithms

- Add BitStringOptimizer for binary optimization with custom JS fitness
- Add PermutationOptimizer for ordering problems (TSP, scheduling)
- Add Nsga2Optimizer for multi-objective optimization with Pareto fronts
- Add custom JS fitness function support to RealVectorOptimizer
- Feature-gate Send+Sync trait bounds for WASM compatibility:
  - Fitness trait in fitness/traits.rs
  - CmaEsFitness trait in algorithms/cmaes.rs
  - MultiObjectiveFitness trait in algorithms/nsga2.rs
  - EvolutionStrategy build/run methods in algorithms/evolution_strategy.rs
- Add new result types: BitStringResult, PermutationResult, MultiObjectiveResult
- All optimizers support custom fitness functions via JS callbacks

Author: claude

crates/fugue-evo-wasm/src/error.rs

@@ -18,7 +18,11 @@ pub fn evolution_error_to_js(err: fugue_evo::error::EvolutionError) -> JsValue {
 
     let obj = js_sys::Object::new();
     let _ = js_sys::Reflect::set(&obj, &"type".into(), &JsValue::from_str(error_type));
-    let _ = js_sys::Reflect::set(&obj, &"message".into(), &JsValue::from_str(&err.to_string()));
+    let _ = js_sys::Reflect::set(
+        &obj,
+        &"message".into(),
+        &JsValue::from_str(&err.to_string()),
+    );
     obj.into()
 }
 

crates/fugue-evo-wasm/src/fitness.rs

@@ -1,5 +1,6 @@
 //! Fitness function support for WASM
 
+use fugue_evo::algorithms::cmaes::CmaEsFitness;
 use fugue_evo::fitness::benchmarks::{Rastrigin, Rosenbrock, Sphere};
 use fugue_evo::fitness::traits::Fitness;
 use fugue_evo::genome::real_vector::RealVector;
@@ -95,3 +96,22 @@ pub fn get_available_fitness_functions() -> Vec<String> {
         "rosenbrock".to_string(),
     ]
 }
+
+/// CMA-ES fitness wrapper (CMA-ES minimizes, so we negate)
+pub struct CmaEsFitnessWrapper {
+    wrapper: FitnessWrapper,
+}
+
+impl CmaEsFitnessWrapper {
+    pub fn new(wrapper: FitnessWrapper) -> Self {
+        Self { wrapper }
+    }
+}
+
+impl CmaEsFitness for CmaEsFitnessWrapper {
+    fn evaluate(&self, x: &RealVector) -> f64 {
+        // CMA-ES minimizes, our FitnessWrapper negates for maximization
+        // So we negate back to get minimization values
+        -self.wrapper.evaluate(x)
+    }
+}