weird result during dataset split when data path starts with `/data`

### Describe the bug

The regex defined here https://github.com/huggingface/datasets/blob/f2607935c4e45c70c44fcb698db0363ca7ba83d4/src/datasets/utils/py_utils.py#L158

will cause a weird result during dataset split when data path starts with `/data`

### Steps to reproduce the bug

1. clone dataset into local path
```
cd /data/train/raw/
git lfs clone https://huggingface.co/datasets/deepmind/code_contests.git

ls /data/train/raw/code_contests
# README.md data dataset_infos.json

ls /data/train/raw/code_contests/data
# test-00000-of-00001-9c49eeff30aacaa8.parquet
# train-[0-9]+-of-[0-9]+-xx.parquet
# valid-00000-of-00001-5e672c5751f060d3.parquet

```

2. loading data from local
```
from datasets import load_dataset
dataset = load_dataset('/data/train/raw/code_contests')

FileNotFoundError: Unable to resolve any data file that matches '['data/train/raw/code_contests/data/train-[0-9][0-9][0-9][0-9][0-9]-of-[0-9][0-9][0-9][0-9][0-9]*.*']' at /data/train/raw/code_contests with any supported extension
```

weird path `data/train/raw/code_contests/data/train-[0-9][0-9][0-9][0-9][0-9]-of-[0-9][0-9][0-9][0-9][0-9]*.*`

While dive deep into `LocalDatasetModuleFactoryWithoutScript` defined in [load.py](https://github.com/huggingface/datasets/blob/f2607935c4e45c70c44fcb698db0363ca7ba83d4/src/datasets/load.py#L627) and _get_data_files_patterns https://github.com/huggingface/datasets/blob/f2607935c4e45c70c44fcb698db0363ca7ba83d4/src/datasets/data_files.py#L228. I found the weird behavior caused by `string_to_dict`

3. check `string_to_dict`
```
p = '/data/train/raw/code_contests/data/test-00000-of-00001-9c49eeff30aacaa8.parquet'
split_pattern = 'data/{split}-[0-9][0-9][0-9][0-9][0-9]-of-[0-9][0-9][0-9][0-9][0-9]*.*'

string_to_dict(p, split_pattern)
# {'split': 'train/raw/code_contests/data/test'}

p = '/data2/train/raw/code_contests/data/test-00000-of-00001-9c49eeff30aacaa8.parquet'
string_to_dict(p, split_pattern)
{'split': 'test'}
```

go deep into string_to_dict https://github.com/huggingface/datasets/blob/f2607935c4e45c70c44fcb698db0363ca7ba83d4/src/datasets/utils/py_utils.py#L158.

4. test the regex:

### Expected behavior

statement in `steps to reproduce the bug`

3. check `string_to_dict`
```
p = '/data/train/raw/code_contests/data/test-00000-of-00001-9c49eeff30aacaa8.parquet'
split_pattern = 'data/{split}-[0-9][0-9][0-9][0-9][0-9]-of-[0-9][0-9][0-9][0-9][0-9]*.*'

string_to_dict(p, split_pattern)
# {'split': 'train/raw/code_contests/data/test'}

p = '/data2/train/raw/code_contests/data/test-00000-of-00001-9c49eeff30aacaa8.parquet'
string_to_dict(p, split_pattern)
{'split': 'test'}
```

### Environment info

- linux(debian)
- python 3.7
- datasets 2.8.0