Dataset with loading script does not work in renamed repos

### Describe the bug

My data repository was first called `BramVanroy/hplt-mono-v1-2` but I then renamed to use underscores instead of dashes. However, it seems that `datasets` retrieves the old repo name when it checks whether the repo contains data loading scripts in this line.

https://github.com/huggingface/datasets/blob/6fb6c834f008996c994b0a86c3808d0a33d44525/src/datasets/load.py#L1845

When I print `filename` it returns `hplt-mono-v1-2.py` but the files in the repo are of course `['.gitattributes', 'README.md', 'hplt_mono_v1_2.py']`. So the `filename` is the original reponame instead of the renamed one.

I am not sure if this is a caching issue or not or how I can resolve it.

### Steps to reproduce the bug

```
from datasets import load_dataset

ds = load_dataset(
"BramVanroy/hplt-mono-v1-2",
"ky",
trust_remote_code=True
)
```

### Expected behavior

That the most recent repo name is used when `filename` is generated.

### Environment info

- `datasets` version: 2.16.1
- Platform: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.34
- Python version: 3.10.13
- `huggingface_hub` version: 0.20.2
- PyArrow version: 14.0.1
- Pandas version: 2.1.3
- `fsspec` version: 2023.10.0