floor_div runtime error for i64, u32 and u64

[TitouanReal]

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

use polars::df;
use polars::prelude::*;

fn main() {
    let df = df![
        "seconds" => [0u64],
    ]
    .unwrap();

    let new_df = df
        .lazy()
        .with_column(col("seconds").floor_div(lit(1u64)))
        .collect()
        .unwrap();

    println!("{new_df:?}");
}

Log output

run StackExec
thread 'main' panicked at /home/titouan/.cargo/registry/src/index.crates.io-6f17d22bba15001f/polars-core-0.41.2/src/series/mod.rs:956:13:
implementation error, cannot get ref UInt64 from Int32

Issue description

The example given above worked in version 0.39.2, but it is broken from 0.40.0 and onwards.

Variations of this error appear when using u64 (given example), u32, and i64, with the error becoming:

• implementation error, cannot get ref UInt64 from Int32,
• implementation error, cannot get ref UInt32 from Int32,
• implementation error, cannot get ref Int64 from Int32,

respectively.

Expected behavior

In version 0.39.2, the example works and outputs :

shape: (1, 1)
┌─────────┐
│ seconds │
│ --- │
│ u64 │
╞═════════╡
│ 0 │
└─────────┘

which is the expected output.

Installed versions

[dependencies.polars]
version = "0.41"
default-features = false
features = ["fmt", "lazy", "round_series"]

[TitouanReal]

A solution to make this work is to use the following code instead:

use polars::df;
use polars::prelude::*;

fn main() {
    let df = df![
        "seconds" => [0u64],
    ]
    .unwrap();

    let new_df = df
        .lazy()
        .with_column((col("seconds") / lit(1u64)).floor().cast(DataType::UInt64))
        .collect()
        .unwrap();

    println!("{new_df:?}");
}

Note that this solution, while working in 0.41, panics in version 0.39.

The performance might be degraded compared to the code proposed in the issue description, but at least it doesn't panic, and it shows the expected behavior.

[TitouanReal]

I can confirm that this panic still occurs in 0.42.0 and 0.43.1.