Parquet writer statistics for categorical/enum values have overall min/max instead of row_group min/max

[deanm0000]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

import polars as pl
import pyarrow.parquet as pq

pl.DataFrame([
    pl.Series('a',['a','b','c'], pl.Categorical)]).write_parquet('cat.parquet',row_group_size=1)
pq.ParquetFile('cat.parquet').metadata.row_group(0).column(0).statistics
<pyarrow._parquet.Statistics object at 0x7ff7545f4680>
  has_min_max: True
  min: a
  max: c # This should be a
  null_count: 0
  distinct_count: None
  num_values: 1 # only 1 value, as expected
  physical_type: BYTE_ARRAY
  logical_type: String
  converted_type (legacy): UTF8

Log output

no stderr

Issue description

When writing a parquet file with a Categorical or Enum type the row group statistics are the overall stats not just what is in the row group. Pyarrow writes the row group statistics correctly

pl.DataFrame([
    pl.Series('a',['a','b','c'], pl.Categorical)]).write_parquet('catpa.parquet',row_group_size=1, use_pyarrow=True)
pq.ParquetFile('catpa.parquet').metadata.row_group(0).column(0).statistics
<pyarrow._parquet.Statistics object at 0x7ff7545b0a90>
  has_min_max: True
  min: a
  max: a # note a here
  null_count: 0
  distinct_count: None
  num_values: 1
  physical_type: BYTE_ARRAY
  logical_type: String
  converted_type (legacy): UTF8

Expected behavior

Polars's parquet writer should write the categorical/enum statistics according to the values not the overall categorical. Without this future predicate pushdowns won't be effective.

Installed versions

<pyarrow._parquet.Statistics object at 0x7ff7545b0a90>
  has_min_max: True
  min: a
  max: a
  null_count: 0
  distinct_count: None
  num_values: 1
  physical_type: BYTE_ARRAY
  logical_type: String
  converted_type (legacy): UTF8

[kszlim]

Would be interesting to see if the Parquet PageIndex is being written appropriately too.