joins break when you rename cols with streaming=True

[marcinplatek]

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

a = pl.LazyFrame({"a": [1, 2, 3]})
b = pl.LazyFrame({"a": [3, 4, 5]})

result = a.join(b, on="a", how="outer_coalesce")

result_broken = a.join(b, on="a", how="outer_coalesce").select(pl.all().name.map(lambda c: c.upper()))

print(result.collect(streaming=False))

print(result.collect(streaming=True))

print(result_broken.collect(streaming=False))

print(result_broken.collect(streaming=True)) # this one for some reason only has rows from b

Log output

shape: (5, 1)
┌─────┐
│ a   │
│ --- │
│ i64 │
╞═════╡
│ 3   │
│ 4   │
│ 5   │
│ 1   │
│ 2   │
└─────┘
shape: (5, 1)
┌─────┐
│ a   │
│ --- │
│ i64 │
╞═════╡
│ 1   │
│ 2   │
│ 3   │
│ 4   │
│ 5   │
└─────┘
shape: (5, 1)
┌─────┐
│ A   │
│ --- │
│ i64 │
╞═════╡
│ 3   │
│ 4   │
│ 5   │
│ 1   │
│ 2   │
└─────┘
shape: (3, 1)
┌─────┐
│ A   │
│ --- │
│ i64 │
╞═════╡
│ 3   │
│ 4   │
│ 5   │
└─────┘

Issue description

As per code, outer_coalesce join between two tables leaves only rows from the second one if a join is followed by a rename of columns and then a streaming collect.

Expected behavior

All should have the same result

Installed versions

--------Version info---------
Polars:               0.20.16
Index type:           UInt32
Platform:             macOS-14.3-arm64-arm-64bit
Python:               3.11.7 (main, Dec  4 2023, 18:10:11) [Clang 15.0.0 (clang-1500.1.0.2.5)]

----Optional dependencies----
adbc_driver_manager:  <not installed>
cloudpickle:          2.2.1
connectorx:           <not installed>
deltalake:            <not installed>
fastexcel:            <not installed>
fsspec:               2023.12.2
gevent:               <not installed>
hvplot:               <not installed>
matplotlib:           <not installed>
numpy:                1.26.4
openpyxl:             <not installed>
pandas:               1.5.3
pyarrow:              15.0.2
pydantic:             2.6.4
pyiceberg:            <not installed>
pyxlsb:               <not installed>
sqlalchemy:           2.0.28
xlsx2csv:             <not installed>
xlsxwriter:           <not installed>```

</details>

[deanm0000]

This worked up to 0.20.13.

0.20.14 is the first version that it is broken.

My test didn't verify that it was explicitly streaming and since collect(streaming=True) automatically will revert to not-streaming when it can't stream that's all that was happening. Since 0.20.14 was the first version that outer joins could stream that means this never worked in streaming not that there was a regression like I originally thought.

[lukemanley]

the behavior on main seems to now be as expected:

>>> print(result.collect(streaming=False))
shape: (5, 1)
┌─────┐
│ a   │
│ --- │
│ i64 │
╞═════╡
│ 3   │
│ 4   │
│ 5   │
│ 1   │
│ 2   │
└─────┘
>>> print(result.collect(streaming=True))
shape: (5, 1)
┌─────┐
│ a   │
│ --- │
│ i64 │
╞═════╡
│ 1   │
│ 2   │
│ 3   │
│ 4   │
│ 5   │
└─────┘
>>> print(result_broken.collect(streaming=False))
shape: (5, 1)
┌─────┐
│ A   │
│ --- │
│ i64 │
╞═════╡
│ 3   │
│ 4   │
│ 5   │
│ 1   │
│ 2   │
└─────┘
>>> print(result_broken.collect(streaming=True))
shape: (5, 1)
┌─────┐
│ A   │
│ --- │
│ i64 │
╞═════╡
│ 2   │
│ 1   │
│ 3   │
│ 4   │
│ 5   │
└─────┘