pyarrow.parquet.read_table("parquet_file") causes bus error in ipython

I have a parquet file with two columns (int64 and double) and 9 million rows. The parquet tools (parquet, parquet-reader, parquet-schema...) read it perfectly. (I have many files, actually, but they all exhibit the same behavior).

The following code fails with "zsh bus error  ipython":

import pyarrow.parquet as pq
pq.read_table("parquet_file")

These snippets work properly.

pq.read_table("parquet_file", use_lagacy_dataset=True)

f = pq.ParquetFile("parquet_file")
f.read()
for batch in f.iterbatches():
print(len(batch))

**Environment**: macOS 12.2.1 aarch64
python. 3.10.1
arrow 7.0.0
**Reporter**: [Jay Edwards](https://issues.apache.org/jira/browse/ARROW-15737)

_{**Note**: *This issue was originally created as [ARROW-15737](https://issues.apache.org/jira/browse/ARROW-15737). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*}

Reproduce the failure with pq.read_table("parquet_file") in IPython on macOS 12.2.1 aarch64, Python 3.10.1, and Arrow 7.0.0, then compare it with the legacy-dataset, ParquetFile.read(), and iterbatches() entry points shown in the report. Done means the default read_table path no longer causes a bus error for the reported two-column, 9-million-row file.

Written by the indexing model from the issue text.