Add arithmetic for arr namespaces

[mcrumiller]

Description

Arrays are equal length series and we should be able to add, subtract, multiply, and divide them if they have the same dtype:

import polars as pl

df = pl.DataFrame({
    "a": pl.Series([[1, 2], [3, 4]], dtype=pl.Array(pl.UInt8, 2)),
    "b": pl.Series([[1, 2], [3, 4]], dtype=pl.Array(pl.UInt8, 2)),
})

# any of these fail, seems reasonable they should be implemented
df["a"] + df["b"]
df.select(pl.col("a") + pl.col("b"))
df.select(pl.sum_horizontal("a", "b"))

pyo3_runtime.PanicException: `add` operation not supported for dtype `array[u8, 2]`

[ritchie46]

Yes, we should support this.

[itamarst]

These all work now (#16791). However, division is broken because the Python code for division tries to cast to Float64, when it should really be casting in this case to pl.Array(pl.Float64, 2).

Building on the example above:

>>> df["a"] / df["b"]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/itamarst/devel/polars/py-polars/polars/series/series.py", line 1082, in __truediv__
    return self.cast(Float64) / other
  File "/home/itamarst/devel/polars/py-polars/polars/series/series.py", line 3992, in cast
    return self._from_pyseries(self._s.cast(dtype, strict, wrap_numerical))
polars.exceptions.InvalidOperationError: cannot cast Array type (inner: 'UInt8', to: 'Float64')

[itamarst]

I will probably fix the division issue as part of #9188, where I have the same problem with List.