'Repo card metadata block was not found' when loading a pragmeval dataset

### Describe the bug

I can't load dataset subsets of 'pragmeval'.

The funny thing is I ran the dataset author's [colab notebook](https://colab.research.google.com/drive/1sg--LF4z7XR1wxAOfp0-3d4J6kQ9nj_A?usp=sharing) and it works just fine. I tried to install exactly the same packages that are installed on colab using poetry, so my environment info only differs from the one from colab in linux version - I still get the same bug outside colab.

### Steps to reproduce the bug

Install dependencies with poetry

pyproject.toml
```
[tool.poetry]
name = "project"
version = "0.1.0"
description = ""
authors = []

[tool.poetry.dependencies]
python = "^3.10"
datasets = "2.16.0"
pandas = "1.5.3"
pyarrow = "10.0.1"
huggingface-hub = "0.19.4"
fsspec = "2023.6.0"

[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"
```

`poetry run python -c "import datasets; print(datasets.get_dataset_config_names('pragmeval'))`

prints ['default']

### Expected behavior

The command should print

```
['emergent',
'emobank-arousal',
'emobank-dominance',
'emobank-valence',
'gum',
'mrda',
'pdtb',
'persuasiveness-claimtype',
'persuasiveness-eloquence',
'persuasiveness-premisetype',
'persuasiveness-relevance',
'persuasiveness-specificity',
'persuasiveness-strength',
'sarcasm',
'squinky-formality',
'squinky-implicature',
'squinky-informativeness',
'stac',
'switchboard',
'verifiability']
```

### Environment info

- `datasets` version: 2.16.0
- Platform: Linux-6.2.0-37-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.19.4
- PyArrow version: 10.0.1
- Pandas version: 1.5.3
- `fsspec` version: 2023.6.0