Update code generation tests with improved sampling

- Implemented top-k sampling (k=50, temperature=0.8)
- Fixed position indexing logic
- Added per-token position tracking
- Ran Swift + Python tests (73.5s total)
- Results: 0 NaN, stable embeddings, but poor code quality
- Issue: Model generates invalid/multilingual characters
- Conclusion: E4B-MarkBase not optimized for code generation
- Recommendation: Use specialized code model for programming tasks
- Test framework: Production-ready, multi-language support

Tests/MarkBaseTests/TestData/TestHelpers.swift

@@ -106,4 +106,27 @@ func verifyOutput(actual: String, expected: String) -> Bool {
     let actualTrimmed = actual.trimmingCharacters(in: .whitespacesAndNewlines)
     let expectedTrimmed = expected.trimmingCharacters(in: .whitespacesAndNewlines)
     return actualTrimmed == expectedTrimmed || actualTrimmed.contains(expectedTrimmed)
+}
+
+func sampleTopK(logits: [Float], k: Int = 40, temperature: Float = 0.9) -> Int {
+    let scaledLogits = logits.map { $0 / temperature }
+    
+    let sorted = scaledLogits.enumerated().sorted { $0.element > $1.element }
+    let topK = sorted.prefix(k)
+    
+    let maxLogit = topK.first?.element ?? 0
+    let probs = topK.map { Float(exp(Double($0.element - maxLogit))) }
+    let sum = probs.reduce(0, +)
+    let normalizedProbs = probs.map { $0 / sum }
+    
+    let random = Float.random(in: 0..<1)
+    var cumulative: Float = 0.0
+    for (index, prob) in normalizedProbs.enumerated() {
+        cumulative += prob
+        if random <= cumulative {
+            return topK[index].offset
+        }
+    }
+    
+    return topK.last?.offset ?? 0
 }