docs(python): correct defaults documented in emw_mean

[MarcoGorelli]

addresses part of #13017

• note the default for adjust in the docstring
• correct the documented default for ignore_nulls

Suppose you have

In [26]: s = pl.Series([5, None, None, 9])

In [27]: alpha = .3

In [28]: w_0 = 1

Then currently, when calculating the result for 9, the weight for 5 is (1-alpha):

In [29]: s.ewm_mean(alpha=alpha)
Out[29]:
shape: (4,)
Series: '' [f64]
[
        5.0
        5.0
        5.0
        7.352941
]

In [30]: w_1 = (1-alpha); print((9*w_0 + 5*w_1)/(w_0+w_1))
7.352941176470589

With ignore_nulls=False, it would be (1-alpha)**3:

In [31]: s.ewm_mean(alpha=alpha, ignore_nulls=False)
Out[31]:
shape: (4,)
Series: '' [f64]
[
        5.0
        5.0
        5.0
        7.978407
]

In [32]: w_1 = (1-alpha)**3; print((9*w_0 + 5*w_1)/(w_0+w_1))
7.978406552494415

Just confirming that the default really is True currently (personally I'd find False more intuitive, but that's for a separate discussion)

[MarcoGorelli]

as per discussion today - this will be deprecated and behaviour brought in-line with the current docstring in 1.0