[BUG] Null types are not promoted in cudf.io.merge_parquet_filemetadata

[rjzamora]

Describe the bug
There is a minor bug in cudf.io.merge_parquet_filemetadata after #14264 was merged. Some details are discussed in that PR. E.g. #14264 (comment)

Steps/Code to reproduce bug

import pyarrow.dataset as ds
import cudf
import fsspec
from dask_cudf.io.parquet import CudfEngine
import pandas as pd
paths = ["/tmp/part.0.parquet", "/tmp/part.1.parquet"]
pd.DataFrame({"a": [None] * 10}).to_parquet(paths[0])
pd.DataFrame({"a": range(10)}).to_parquet(paths[1])
md1 = CudfEngine.collect_file_metadata(paths[0], fsspec, "none")
md2 = CudfEngine.collect_file_metadata(paths[1], fsspec, "none")
md = cudf.io.merge_parquet_filemetadata([md1, md2])
with open("/tmp/_metadata", "wb") as fil:
    fil.write(memoryview(md))
schema = ds.parquet_dataset("/tmp/_metadata").schema
assert schema.field(0).type !=  “null”

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
Cell In[3], line 15
     13     fil.write(memoryview(md))
     14 schema = ds.parquet_dataset("/tmp/_metadata").schema
---> 15 assert schema.field(0).type != "null"

AssertionError: 

Expected behavior
Null column metadata should be "promoted" when it is merged with integer metedata for the same column.

Additional context
This bug is unlikely to affect many users, and can be treated as a low priority if the fix proves difficult.