feat: add benchmark suite for pipeline operations

- Add pytest-benchmark to dev dependencies
- Benchmark SVD decomposition on synthetic vote matrix
- Benchmark cosine similarity at small/medium/large scales

P5-003: Benchmark suite

tests/benchmark/test_pipeline_benchmarks.py

@@ -0,0 +1,54 @@
+import numpy as np
+import pytest
+from sklearn.decomposition import TruncatedSVD
+from sklearn.metrics.pairwise import cosine_similarity
+
+
+class TestSVDBenchmark:
+    @pytest.mark.benchmark
+    def test_svd_on_synthetic_vote_matrix(self, benchmark):
+        """Benchmark SVD decomposition on a 100x20 synthetic vote matrix."""
+        np.random.seed(42)
+        vote_matrix = np.random.choice([-1, 0, 1], size=(100, 20))
+
+        def run_svd():
+            svd = TruncatedSVD(n_components=5, random_state=42)
+            return svd.fit_transform(vote_matrix)
+
+        result = benchmark(run_svd)
+        assert result.shape == (100, 5)
+
+
+class TestSimilarityBenchmark:
+    @pytest.mark.benchmark
+    def test_cosine_similarity_small(self, benchmark):
+        """Benchmark cosine similarity on 50 vectors of dimension 10."""
+        vectors = np.random.randn(50, 10).astype(np.float32)
+
+        def run_similarity():
+            return cosine_similarity(vectors)
+
+        result = benchmark(run_similarity)
+        assert result.shape == (50, 50)
+
+    @pytest.mark.benchmark
+    def test_cosine_similarity_medium(self, benchmark):
+        """Benchmark cosine similarity on 200 vectors of dimension 50."""
+        vectors = np.random.randn(200, 50).astype(np.float32)
+
+        def run_similarity():
+            return cosine_similarity(vectors)
+
+        result = benchmark(run_similarity)
+        assert result.shape == (200, 200)
+
+    @pytest.mark.benchmark
+    def test_cosine_similarity_large(self, benchmark):
+        """Benchmark cosine similarity on 500 vectors of dimension 100."""
+        vectors = np.random.randn(500, 100).astype(np.float32)
+
+        def run_similarity():
+            return cosine_similarity(vectors)
+
+        result = benchmark(run_similarity)
+        assert result.shape == (500, 500)