ewm_mean default value does not match docstring and pandas API

[PrettyMeng]

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
df = pl.DataFrame({
    "values": [1, 2, 3, None, 5]
})
alpha = 0.5
df = df.with_columns([
    df["values"].ewm_mean(alpha=alpha).alias("ewm_mean")
])

Log output

No response

Issue description

The default value of ignore_nulls does not match docstring: "When ignore_nulls=False (default), weights are based on absolute positions." . But it actually defaults to True. Also this is inconsistent with pandas API.

Two solutions I came up with:

• Simply change the docstring
• Change the default value to False as in fix(python): change ignore_nulls default to False #13004

I prefer the second one to keep the same interface as in pandas, which could make code migration from pandas to polars easier.

Expected behavior

ignore_nulls should default to False, or at least be consistent with doc string

Installed versions

--------Version info---------
Polars:      0.18.4
Index type:  UInt32
Platform:    macOS-10.16-x86_64-arm64-64bit
Python:      3.8.0 (default, Nov  6 2019, 15:49:01)
[Clang 4.0.1 (tags/RELEASE_401/final)]

----Optional dependencies----
numpy:       1.24.4
pandas:      2.0.3
pyarrow:     14.0.1
connectorx:  0.3.2
deltalake:   0.14.0
fsspec:      2023.12.2
matplotlib:  3.7.4
xlsx2csv:    0.8.1
xlsxwriter:  3.1.9

[ritchie46]

@MarcoGorelli any opinion here?

[MarcoGorelli]

I'm no ewm expert, I just think I'd find ignore_nulls=False a more intuitive default

@alexander-beedie @matteosantama do you have an opinion on this one?

[MarcoGorelli]

as per discussion today - behaviour will be brought in-line with the current docstring via deprecation which will be enforced in 1.0