fix(dedup): use deterministic seed in TestChunker_ShiftedData

The test was flaky because it used crypto/rand for random data,
causing non-deterministic chunk boundaries. With small sample sizes
(100KB / 8KB avg = ~12 chunks), variance was high - sometimes only
42.9% overlap instead of expected >50%.

Fixed by using math/rand with seed 42 for reproducible test results.
Now consistently achieves 91.7% overlap (11/12 chunks).

internal/dedup/chunker_test.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"crypto/rand"
 	"io"
+	mathrand "math/rand"
 	"testing"
 )
 
@@ -100,12 +101,15 @@ func TestChunker_Deterministic(t *testing.T) {
 
 func TestChunker_ShiftedData(t *testing.T) {
 	// Test that shifted data still shares chunks (the key CDC benefit)
+	// Use deterministic random data for reproducible test results
+	rng := mathrand.New(mathrand.NewSource(42))
+	
 	original := make([]byte, 100*1024)
-	rand.Read(original)
+	rng.Read(original)
 
 	// Create shifted version (prepend some bytes)
 	prefix := make([]byte, 1000)
-	rand.Read(prefix)
+	rng.Read(prefix)
 	shifted := append(prefix, original...)
 
 	// Chunk both