The error message "is not a Parquet file" doesn't show the Parquet path anymore

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

Version 1.13.1, used by Spark 3.5: https://github.com/apache/parquet-java/blob/db4183109d5b734ec5930d870cdae161e408ddba/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java#L548

used to show the parquet file path: `Cause: java.lang.RuntimeException: file:/tmp/spark-37..../00000000000000000000.checkpoint.parquet is not a Parquet file (length is too low: 10)`

In version 1.15.2, used by Spark 4: https://github.com/apache/parquet-java/blob/859eac165b08f927fa14590c33bc5f476405fb68/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java#L584

The error message says: `Cause: java.lang.RuntimeException: io.delta.kernel.defaults.internal.parquet.ParquetIOUtils$1@63d4ffeb is not a Parquet file (length is too low: 10)`

### Component(s)

_No response_

Start by comparing the linked ParquetFileReader.java locations from versions 1.13.1 and 1.15.2, focusing on how the error message obtains the path. Reproduce the reported short-file case using the shown parquet path and verify that the resulting exception identifies that path instead of the ParquetIOUtils object representation.

Written by the indexing model from the issue text.