TestParquetReader fails with hadoop 3.4.2

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

A couple of test cases of `TestParquetReader` suite started failing with errors like:
```
- testRangeFiltering[0] *** FAILED ***
org.apache.parquet.bytes.TrackingByteBufferAllocator$LeakedByteBufferException: 24 ByteBuffer object(s) is/are remained unreleased after closing this allocator.
at org.apache.parquet.bytes.TrackingByteBufferAllocator.close(TrackingByteBufferAllocator.java:161)
at org.apache.parquet.hadoop.TestParquetReader.closeAllocator(TestParquetReader.java:175)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.base/java.lang.reflect.Method.invoke(Method.java:569)
```
with hadoop 3.4.2 and parquet-mr 1.16.0
The leaks are happening when reading using vectored IO because we never pass the buffers to the releaser.

### Component(s)

_No response_

Start with the failing TestParquetReader cases, especially testRangeFiltering, and inspect the vectored IO reading path. Check how returned buffers are handled by the releaser, then rerun the suite with Hadoop 3.4.2 and parquet-mr 1.16.0. Done means the tests pass without TrackingByteBufferAllocator leak errors.

Written by the indexing model from the issue text.