Median doesn't have same precision as min/max in datetime

[etiennebacher]

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

from datetime import datetime
import polars as pl

dat = pl.DataFrame(data={"datetime": [datetime.max]})

dat.select(
    min=pl.col("datetime").min(), 
    max=pl.col("datetime").max(),
    median=pl.col("datetime").median()
)

Log output

shape: (1, 3)
┌─────────────────┬─────────────────┬───────────────────────┐
│ min             ┆ max             ┆ median                │
│ ---             ┆ ---             ┆ ---                   │
│ datetime[μs]    ┆ datetime[μs]    ┆ datetime[μs]          │
╞═════════════════╪═════════════════╪═══════════════════════╡
│ 9999-12-31      ┆ 9999-12-31      ┆ +10000-01-01 00:00:00 │
│ 23:59:59.999999 ┆ 23:59:59.999999 ┆                       │
└─────────────────┴─────────────────┴───────────────────────┘

Issue description

Median is different from the others while it should be equal. This is also a bit related to:

• Series.quantile returns None with datetime type data #12448: .quantile() doesn't work on datetime (returns null) but median works
• OverflowError when running DataFrame.describe() and column contains datetime.max value #14573: .describe() panicks with datetime.max because of overflow (likely because median is > 10000)

Expected behavior

Min, median and max should all have the same value in this case.

Installed versions

--------Version info---------
Polars:              1.20.0
Index type:          UInt32
Platform:            Linux-6.8.0-50-generic-x86_64-with-glibc2.39
Python:              3.12.3 (main, Nov  6 2024, 18:32:19) [GCC 13.2.0]
LTS CPU:             False

----Optional dependencies----
Azure CLI            <not installed>
adbc_driver_manager  1.3.0
altair               5.5.0
azure.identity       <not installed>
boto3                1.35.88
cloudpickle          3.1.0
connectorx           0.4.0
deltalake            0.22.3
fastexcel            0.12.0
fsspec               2024.12.0
gevent               24.11.1
google.auth          2.37.0
great_tables         0.15.0
matplotlib           3.10.0
numpy                2.0.2
openpyxl             3.1.5
pandas               2.2.3
pyarrow              18.1.0
pydantic             2.10.4
pyiceberg            <not installed>
sqlalchemy           2.0.36
torch                <not installed>
xlsx2csv             0.8.4
xlsxwriter           3.2.0