Hive partitions are corrupted during reads from cloud storage in Polars 1.0.0-rc.1

[nameexhaustion]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

import polars as pl
import os
from polars.testing import assert_frame_equal
from pathlib import Path

os.environ["POLARS_FORCE_ASYNC"] = "1"
os.environ["POLARS_PREFETCH_SIZE"] = "1"

root = Path(".env/data2")

dfs = [
    pl.DataFrame({"x": 1}),
    pl.DataFrame({"x": 2}),
    pl.DataFrame({"x": 3}),
]

paths = [
    root / "a=1/b=1/data.bin",
    root / "a=2/b=2/data.bin",
    root / "a=3/b=3/data.bin",
]

[
    [paths[i].parent.mkdir(exist_ok=True, parents=True), dfs[i].write_parquet(paths[i])]
    for i in range(len(dfs))
]

lf = pl.scan_parquet(root)

assert_frame_equal(lf.collect(), pl.DataFrame({k: [1, 2, 3] for k in ["x", "a", "b"]}))

Output:

AssertionError: DataFrames are different (value mismatch for column 'a')
[left]:  [1, 1, 1]
[right]: [1, 2, 3]

Log output

No response

Issue description

This is introduced by 306a918. It will occur when reading more files than the calculated prefetch size.

Expected behavior

Given example passes

Installed versions

main @ 46ba436