df.drop(cs.ends_with()) returning ColumnNotFoundError when columns exist in 1.4.0 (worked in 1.3.0)

[ATL2001]

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 os
os.environ["POLARS_VERBOSE"] = "1"

import polars as pl
from polars import selectors as cs 

print(pl.__version__)
df = pl.DataFrame([["x"], [1]], schema={"foo":pl.String, "foo_right":pl.Int8})
print(df)
print(df.drop(
    cs.ends_with("_right")
))

Log output

ColumnNotFoundError: "^.*(_right)$" not found

Issue description

version 1.4.0 is telling me it can't find a column that exists when I use cs.ends_with() the same code produces the expected result in 1.3.0

Expected behavior

in version 1.3.0 the same code resulted in the expected result of just the "foo" column

1.3.0
shape: (1, 2)
+-----+-----------+
| foo | foo_right |
| --- | ---       |
| str | i8        |
+=================+
| x   | 1         |
+-----+-----------+
shape: (1, 1)
+-----+
| foo |
| --- |
| str |
+=====+
| x   |
+-----+

Installed versions

--------Version info---------
Polars:               1.4.0
Index type:           UInt32
Platform:             Windows-10-10.0.19045-SP0
Python:               3.9.16 [MSC v.1931 64 bit (AMD64)]

----Optional dependencies----
adbc_driver_manager:  <not installed>
cloudpickle:          2.0.0
connectorx:           0.3.2
deltalake:            <not installed>
fastexcel:            <not installed>
fsspec:               2022.7.1
gevent:               <not installed>
great_tables:         <not installed>
hvplot:               <not installed>
matplotlib:           3.6.0
nest_asyncio:         1.5.5
numpy:                1.20.1
openpyxl:             3.0.10
pandas:               1.4.4
pyarrow:              1.0.1
pydantic:             2.7.1
pyiceberg:            <not installed>
sqlalchemy:           1.4.39
torch:                <not installed>
xlsx2csv:             0.8.2
xlsxwriter:           3.2.0

[cmdlineluser]

It seems to be specific to .drop

>>> df.select(~cs.ends_with("_right"))
shape: (1, 1)
┌─────┐
│ foo │
│ --- │
│ str │
╞═════╡
│ x   │
└─────┘

There was a .drop fix pushed for #18008 earlier which may be related?

• perf: Delay selection expansion #18011

---

Update: Just confirmed that it worked up to that commit.

Perhaps @ritchie46 can take a look.

[ATL2001]

good call, I updated the title with the df.drop()

[kszlim]

I'm experiencing this too, affects cs.starts_with and cs.contains as well

[ATL2001]

As always, Thanks Ritchie!