Why is group_by() output always identically sorted?

Moved from: https://github.com/eitsupi/neo-r-polars/issues/339

---

Python polars and old r-polars return dataframes with different row order, but the rewritten polars returns all dataframes identically sorted:

import polars as pl

n_obs = 6
df = pl.LazyFrame({
    "feature": ["b", "b", "c", "c", "a", "a"],
    "value": list(range(n_obs)),
})

for i in range(10):
    print(
        df.group_by("feature").agg([
            pl.col("value").mean(),
        ]).collect()
    )

df = pl$LazyFrame(
    feature = c("b", "b", "c", "c", "a", "a"),
    value = 1:6,
)

for (i in 1:10) {
  print(
    df$group_by("feature")$agg(
      pl$len(),
      pl$col("value")$mean(),
    )$collect()
  )
}