fix: fixing unnesscary raise of mean_horizontal

[Cheukting]

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

• Related issue bug: nw.mean_horizontal(nw.all()) raises unnecessarily #1081
• Closes bug: nw.mean_horizontal(nw.all()) raises unnecessarily #1081

Checklist

• Code follows style guide (ruff)
• Tests added
• Documented the changes

If you have comments or can explain your changes, please do so below.

See #1081

[Cheukting]

mean_horizontal was implemented in Polars 0.20.8: https://github.com/pola-rs/polars/releases/tag/py-0.20.8

I found the implementation in Narwhals and it seems need an update as well, will do that in a commit

[FBruzzesi]

Hey @Cheukting thanks for addressing this!

Probably it's an obsession of mine, but can we keep using generator comprehensions instead of list and for loops?

[Cheukting]

All checks have passed @MarcoGorelli ship it! (What cute animal do we have time?)

@FBruzzesi are you sure you want me to refactor them? Because in the for loop there is already a list comprehension, it will become a list comprehension within a list comprehension... it will not be very readable (IMHO)... here is a preview example:

            series = [
                series.extend([_series.fill_null(0) for _series in _expr._call(df)]) for _expr in parsed_exprs
            ]

I can refactor with no problem if it is preferred. Just wanna double check.

[FBruzzesi]

Hey hey @Cheukting

All checks have passed @MarcoGorelli ship it! (What cute animal do we have time?)

Awesome! I will let Marco pick animal(s) 😂

are you sure you want me to refactor them? Because in the for loop there is already a list comprehension, it will become a list comprehension within a list comprehension... it will not be very readable (IMHO)

I did the same in #1090:

series = (s for _expr in parsed_exprs for s in _expr._call(df))

I know it is nested, but I believe it is still very readable for anyone who has some python experience.

Actually I think this PR should take #1090 into account and set

- output_names=parsed_exprs[0]._output_names,
+ output_names=reduce_output_names(parsed_exprs),

in mean_horizontal as well, otherwise we can end up in a similar issue to what's addressed there

[MarcoGorelli]

Thanks @Cheukting , and thanks @FBruzzesi for reviewing!

love this