[Java] ArrowRecordBatch Body Length Computation Only Accounts for Last Buffer

### Describe the bug, including details regarding any error messages, version, and platform.

Hi Arrow Community,

I believe `ArrowRecordBatch::computeBodyLength` has an implementation bug. It does the following

```
for (int i = 0; i < buffers.size(); i++) {
ArrowBuf buffer = buffers.get(i);
ArrowBuffer layout = buffersLayout.get(i);
size = layout.getOffset() + buffer.readableBytes();

// round up size to the next multiple of 8
size = DataSizeRoundingUtil.roundUpTo8Multiple(size);
}
return size;
```

Since `size` is being assigned and not incremented at first - does this mean computeBodyLength only accounts for the last buffer ?

### Component(s)

Java

从 ArrowRecordBatch::computeBodyLength 开始，检查缓冲区布局和可读字节数如何影响返回值。使用多个缓冲区复现该计算，然后检查周围的 Java 代码或测试，以确认预期的正文长度语义。完成的标准是：计算涵盖每个缓冲区，并保留所需的 8 字节对齐。

由索引模型根据 Issue 内容生成。