feat: add ewm_mean_by

[MarcoGorelli]

closes #10794

Demo of why this matters: for non-evenly spaced points, there's a noticeable difference compared with plain ewm_mean:

---

question - should it sort on behalf of users, or raise?

What I've tried introducing here is a check_sorted argument:

• if it's set to check_sorted=False, then Polars assumes your data to be
• else, instead of erroring, Polars takes care of this for you

Trying to follow @orlp 's guidance here

Other options could be:

• if the data isn't sorted, then sort for the user, and raise a warning. What's annoying about this one is that sortedness flags aren't preserved in .over, and so pl.col('a').ewm_mean_by('times', half_life='4d').over('groups') would always warn. I'd find that quite annoying as a user
• raise if the data isn't sorted. As a user, I find such errors a bit annoying, but not terribly so

[codspeed-hq]

CodSpeed Performance Report

Merging #15638 will not alter performance

_{Comparing MarcoGorelli:ewm-mean-time (2d0b52c) with main (8ef2e21)}

Summary

✅ 22 untouched benchmarks

[codecov]

Codecov Report

Attention: Patch coverage is 98.46154% with 3 lines in your changes are missing coverage. Please review.

❗ No coverage uploaded for pull request base (main@19f0939). Click here to learn what that means.
Report is 1 commits behind head on main.

❗ Current head ad8bb68 differs from pull request most recent head 2d0b52c. Consider uploading reports for the commit 2d0b52c to get more accurate results

Files	Patch %	Lines
crates/polars-ops/src/series/ops/ewm_by.rs	98.52%	2 Missing ⚠️
crates/polars-plan/src/dsl/function_expr/mod.rs	83.33%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #15638   +/-   ##
=======================================
  Coverage        ?   81.34%           
=======================================
  Files           ?     1374           
  Lines           ?   176473           
  Branches        ?     2544           
=======================================
  Hits            ?   143549           
  Misses          ?    32443           
  Partials        ?      481           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ritchie46]

nit: We could do the assert immediately in polars_ensure

[hixan]

I know I am a little late to the party, but would it be a better approach to give the ability to provide the half_life as an expression instead? This would allow the above behaviour with something like: (taking liberties with the half life dtype)

def ewm_mean_by(input_column, times_column, alpha):
    return input_column.ewm_mean(alpha = alpha * times_column.diff())

(haven't checked the math here), It could be used to perform the operation in update<T> in this PR while giving users the ability to also define variable half lives

[MarcoGorelli]

could you open a new issue for this please?