feat: v0.2.0 AriaEngine 自研存储引擎

- 新增 AriaEngine: LSM-Tree 页面式存储引擎，19 个模块，~3500 行 TS
  - page/: Slotted Page 格式 (header/slot/tuple/format) + CRC32
  - buffer/: Buffer Pool (LRU 缓存 + 驱逐策略)
  - index/: LSM-Tree (MemTable 红黑树 + SSTable + Bloom Filter + Merge Iterator)
  - wal/: WAL 日志 (二进制格式) + Checkpoint 管理
  - transaction/: MVCC 版本链 + 快照隔离
  - store/: IndexedDB / Memory 双后端抽象
  - compression/: LZ4 页面压缩

- 完整持久化: Schema 自动保存、SSTable 元数据管理、WAL 恢复
- 事务感知 CRUD: insert/update/delete 在事务中缓冲到 snapshot
- mode: 'aria' 激活自研引擎

- 新增 7 个测试文件，测试数 318 → 524，套件 20 → 27
  - aria-page.test.ts (32 tests): Page 格式单元测试
  - aria-index.test.ts (26 tests): Bloom Filter + MemTable
  - aria-sstable.test.ts (9 tests): SSTable Builder + Reader
  - aria-buffer.test.ts (25 tests): LRU + Eviction + Buffer Pool
  - aria-wal-mvcc.test.ts (22 tests): WAL 编解码 + MVCC 事务
  - aria-compress.test.ts (11 tests): LZ4 + Merge Iterator
  - aria.test.ts (80 tests): AriaEngine 集成 + 边界测试

- Bug 修复: LRUList size 跟踪、WAL 缓冲区越界、ColumnEncoding 导入
- 全面更新 README.md + site/ 站点文件 (index/docs/demo)

tests/engine/aria-compress.test.ts

@@ -0,0 +1,123 @@
+/**
+ * AriaEngine LZ4 压缩 + LSM Merge Iterator 单元测试
+ */
+import { compressLZ4, decompressLZ4 } from '../../src/engine/aria/compression/lz4';
+import { MergeIterator, ArrayEntrySource } from '../../src/engine/aria/index/merge_iterator';
+
+// ===================================================================
+//  LZ4 压缩
+// ===================================================================
+describe('AriaEngine — LZ4 Compression', () => {
+  it('压缩+解压往返 — 简单文本', () => {
+    const input = new TextEncoder().encode('hello world hello world hello world');
+    const compressed = compressLZ4(input);
+    const decompressed = decompressLZ4(compressed, input.byteLength);
+    expect(Array.from(decompressed)).toEqual(Array.from(input));
+  });
+
+  it('压缩+解压 — 重复数据', () => {
+    const pattern = 'ABCD';
+    const repeated = pattern.repeat(100);
+    const input = new TextEncoder().encode(repeated);
+    const compressed = compressLZ4(input);
+    // 重复数据应该有较好压缩率
+    expect(compressed.byteLength).toBeLessThan(input.byteLength);
+    const decompressed = decompressLZ4(compressed, input.byteLength);
+    expect(new TextDecoder().decode(decompressed)).toBe(repeated);
+  });
+
+  it('压缩 — 太短不压缩', () => {
+    const input = new Uint8Array([1, 2]);
+    const compressed = compressLZ4(input);
+    expect(compressed.byteLength).toBe(input.byteLength);
+  });
+
+  it('压缩 — 不可压缩数据返回原始', () => {
+    // 随机数据是不可压缩的
+    const input = new Uint8Array(256);
+    for (let i = 0; i < 256; i++) input[i] = Math.floor(Math.random() * 256);
+    const compressed = compressLZ4(input);
+    // 不可压缩时应返回原始大小或更小
+    expect(compressed.byteLength).toBeGreaterThanOrEqual(0);
+  });
+
+  it('解压 — 恢复原始数据', () => {
+    const input = new TextEncoder().encode('The quick brown fox jumps over the lazy dog. '.repeat(10));
+    const compressed = compressLZ4(input);
+    const decompressed = decompressLZ4(compressed, input.byteLength);
+    expect(new TextDecoder().decode(decompressed)).toBe('The quick brown fox jumps over the lazy dog. '.repeat(10));
+  });
+
+  it('压缩后大小不超过原始', () => {
+    for (const size of [10, 50, 100, 200, 500]) {
+      const input = new Uint8Array(size);
+      for (let i = 0; i < size; i++) input[i] = i % 256;
+      const compressed = compressLZ4(input);
+      expect(compressed.byteLength).toBeLessThanOrEqual(input.byteLength + 16);
+    }
+  });
+});
+
+// ===================================================================
+//  MergeIterator
+// ===================================================================
+describe('AriaEngine — MergeIterator', () => {
+  it('单数据源归并', () => {
+    const mi = new MergeIterator();
+    mi.addSource(new ArrayEntrySource([
+      ['a', { v: 1 }],
+      ['b', { v: 2 }],
+      ['c', { v: 3 }],
+    ]));
+    const result = mi.drain();
+    expect(result).toHaveLength(3);
+    expect(result.map(([k]) => k)).toEqual(['a', 'b', 'c']);
+  });
+
+  it('多数据源归并去重（保留最新）', () => {
+    const mi = new MergeIterator();
+    mi.addSource(new ArrayEntrySource([['a', { v: 'new' }], ['c', { v: 3 }]]));
+    mi.addSource(new ArrayEntrySource([['a', { v: 'old' }], ['b', { v: 2 }]]));
+    const result = mi.drain();
+    expect(result).toHaveLength(3);
+    expect(result[0][0]).toBe('a');
+    expect(result[0][1].v).toBe('new');
+    expect(result[1][0]).toBe('b');
+    expect(result[2][0]).toBe('c');
+  });
+
+  it('空数据源归并', () => {
+    const mi = new MergeIterator();
+    mi.addSource(new ArrayEntrySource([]));
+    const result = mi.drain();
+    expect(result).toHaveLength(0);
+  });
+
+  it('大数量归并', () => {
+    const mi = new MergeIterator();
+    for (let s = 0; s < 5; s++) {
+      const entries: [string, Record<string, unknown>][] = [];
+      for (let i = 0; i < 100; i++) {
+        entries.push([`src${s}-key-${String(i).padStart(3, '0')}`, { src: s, idx: i }]);
+      }
+      mi.addSource(new ArrayEntrySource(entries));
+    }
+    const result = mi.drain();
+    expect(result).toHaveLength(100);
+  });
+
+  it('ArrayEntrySource — 迭代器用完返回 null', () => {
+    const src = new ArrayEntrySource([['k', { v: 1 }]]);
+    expect(src.next()).not.toBeNull();
+    expect(src.next()).toBeNull();
+    expect(src.next()).toBeNull();
+  });
+
+  it('ArrayEntrySource — reset 重置', () => {
+    const src = new ArrayEntrySource([['k1', { v: 1 }], ['k2', { v: 2 }]]);
+    src.next();
+    src.reset();
+    const val = src.next();
+    expect(val![0]).toBe('k1');
+  });
+});