docs: Make page about pandas null handling

[MarcoGorelli]

We sometimes reference pandas and Polars handling null values differently in docs, but it's a bit scattered - I think we should:

• only include it in the docstring if the function specifically modifies / counts/ drops null values (e.g. fill_null, null_count)
• reference the page in those cases

I'm not sure we should be including Polars examples which includes 'nan' - the reason is that 'nan' is quite rare to encounter in Polars. Missing values are represented by null, and you'd only get 'nan' if you did something "illegal" like divide 0 by 0

My concern is that otherwise we distract from the main behaviour of the function in question

[MarcoGorelli]

all_horizontal probably needs updating https://narwhals-dev.github.io/narwhals/api-reference/narwhals/#narwhals.all_horizontal, as it currently just says pandas and Polars handle null values differently.

and the example shows pandas with object dtype, whereas I think we should showcase boolean columns