Buffers allocated by ParquetFileReader.readVectored() are not being released

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

this surfaces when running `TestParquetReader` with hadoop >= 3.4.0
```
mvn -Dtest=TestParquetReader -Dhadoop.version=3.4.1
```

This is disappointing as it means
1. the hadoop release process hasn't been doing the parquet build and test (it's an optional action)
2. this got through our own large scale QE testing.

The good news: this hasn't surfaced in production, presumably because nothing is trying to read into direct buffers -something parquet blocks automatically.

```

[ERROR] org.apache.parquet.hadoop.TestParquetReader.testCurrentRowIndex[2] -- Time elapsed: 0.013 s <<< ERROR!
org.apache.parquet.bytes.TrackingByteBufferAllocator$LeakedByteBufferException: 2 ByteBuffer object(s) is/are remained unreleased after closing this allocator.
at org.apache.parquet.bytes.TrackingByteBufferAllocator.close(TrackingByteBufferAllocator.java:171)
at org.apache.parquet.hadoop.TestParquetReader.closeAllocator(TestParquetReader.java:175)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.junit.runners.model.FrameworkMethod$1.runReflectiveCall(FrameworkMethod.java:59)
at org.junit.internal.runners.model.ReflectiveCallable.run(ReflectiveCallable.java:12)
Caused by: org.apache.parquet.bytes.TrackingByteBufferAllocator$ByteBufferAllocationStacktraceException: Allocation stacktrace of the first ByteBuffer:
at org.apache.parquet.bytes.TrackingByteBufferAllocator$ByteBufferAllocationStacktraceException.create(TrackingByteBufferAllocator.java:100)
at org.apache.parquet.bytes.TrackingByteBufferAllocator$ByteBufferAllocationStacktraceException.access$300(TrackingByteBufferAllocator.java:94)
at org.apache.parquet.bytes.TrackingByteBufferAllocator.allocate(TrackingByteBufferAllocator.java:140)
at org.apache.hadoop.fs.RawLocalFileSystem$LocalFSFileInputStream.readVectored(RawLocalFileSystem.java:336)
at org.apache.hadoop.fs.BufferedFSInputStream.readVectored(BufferedFSInputStream.java:182)
at org.apache.hadoop.fs.FSDataInputStream.readVectored(FSDataInputStream.java:307)
at org.apache.hadoop.fs.ChecksumFileSystem$ChecksumFSInputChecker.readVectored(ChecksumFileSystem.java:444)
at org.apache.hadoop.fs.FSDataInputStream.readVectored(FSDataInputStream.java:307)
```

### Component(s)

Core

先从 TestParquetReader.java 中 closeAllocator 附近以及 readVectored 的堆栈跟踪开始，然后检查 ParquetFileReader.readVectored() 路径和 TrackingByteBufferAllocator 的使用情况。使用 `mvn -Dtest=TestParquetReader -Dhadoop.version=3.4.1` 复现；完成标准是在 Hadoop 3.4.0 或更高版本下测试通过且没有泄漏的缓冲区。

由索引模型根据 Issue 内容生成。