perf: compose 30% faster — move semantics, BFS offsets, zero-alloc hex formatting

- compose.cpp: emitLine takes LineMeta&& (move, not copy) at all 22 call sites
- compose.cpp: reserve meta/text buffers, BFS offset computation O(N) vs O(N*D)
- compose.cpp: pre-compute typeNameLens[], merge global width loops
- format.cpp: bytesToHex uses stack buffer + lookup table (zero heap allocs)
- format.cpp: hexVal single QString::asprintf instead of 2-string concat
- editor.cpp: guard hover updates during applyDocument (stale index safety)
- core.h: assertion on makeArrayElemSelId negative index
- format.cpp: assertion on extractBits overflow
- main.cpp: tree lines enabled by default
- bench_large_class: add 2000-field benchComposeLarge test

Benchmark: 500 fields 0.70→0.51ms (27%), 2000 fields 2.28→1.57ms (31%)

tests/bench_large_class.cpp

@@ -65,6 +65,7 @@ private:
 private slots:
     void initTestCase();
     void benchCompose();
+    void benchComposeLarge();
     void benchApplyDocument();
     void benchHoverHighlight();
     void benchSelectionOverlay();
@@ -112,6 +113,36 @@ void BenchLargeClass::benchCompose()
     QVERIFY(elapsed > 0);
 }
 
+void BenchLargeClass::benchComposeLarge()
+{
+    // Build a 2000-field tree to stress-test compose at scale
+    NodeTree bigTree = buildLargeTree(2000);
+    QByteArray buf(0x40000, '\0');
+    for (int i = 0; i < buf.size(); ++i) buf[i] = (char)(i & 0xFF);
+    BufferProvider bigProv(buf, QStringLiteral("bench_large"));
+
+    // Warmup
+    { ComposeResult w = rcx::compose(bigTree, bigProv); Q_UNUSED(w); }
+
+    const int ITERS = 50;
+    QElapsedTimer timer;
+
+    timer.start();
+    for (int i = 0; i < ITERS; ++i) {
+        ComposeResult r = rcx::compose(bigTree, bigProv);
+        Q_UNUSED(r);
+    }
+    qint64 elapsed = timer.elapsed();
+
+    qDebug() << "";
+    qDebug() << "=== Compose Benchmark (2000 fields) ===";
+    qDebug() << "  Tree:" << bigTree.nodes.size() << "nodes";
+    qDebug() << "  Iterations:" << ITERS;
+    qDebug() << "  Total:" << elapsed << "ms";
+    qDebug() << "  Per-compose:" << (double)elapsed / ITERS << "ms";
+    QVERIFY(elapsed > 0);
+}
+
 void BenchLargeClass::benchApplyDocument()
 {
     RcxEditor editor;