Remove config names as yaml keys

[lhoestq]

Many datasets have dots in their config names. However it causes issues with the YAML tags of the dataset cards since we can't have dots in YAML keys.

I fix this, I removed the tags separations per config name completely, and have a single flat YAML for all configurations. Dataset search doesn't use this info anyway. I removed all the config names used as YAML keys, and I moved them in under a new config: key.

This is related to #2362 (internal https://github.com/huggingface/moon-landing/issues/946).

Also removing the dots in the YAML keys would allow us to do as in #4302 which removes a hack that replaces all the dots by underscores in the YAML tags.

I also added a test in the CI that checks that all the YAML tags to make sure that:

• they can be parsed using a YAML parser
• they contain only valid YAML tags like languages or task_ids

[lhoestq]

I included the change from #4302 directly in this PR, this way the datasets will be updated right away in the CI (the CI is only triggered when a dataset card is changed)

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[lhoestq]

Alright it's ready now :)

Here is an example for the ade_corpus_v2 dataset card. Notice the new configs key:

datasets/datasets/ade_corpus_v2/README.md

Lines 1 to 78 in 76d9a14

	---
	annotations_creators:
	- expert-generated
	language_creators:
	- found
	languages:
	- en
	licenses:
	- unknown
	multilinguality:
	- monolingual
	size_categories:
	- 10K<n<100K
	- 1K<n<10K
	- n<1K
	source_datasets:
	- original
	task_categories:
	- text-classification
	- token-classification
	task_ids:
	- coreference-resolution
	- fact-checking
	paperswithcode_id: null
	pretty_name: Adverse Drug Reaction Data v2
	train-eval-index:
	- config: Ade_corpus_v2_classification
	task: text-classification
	task_id: multi_class_classification
	splits:
	train_split: train
	col_mapping:
	text: text
	label: target
	metrics:
	- type: accuracy
	name: Accuracy
	- type: f1
	name: F1 macro
	args:
	average: macro
	- type: f1
	name: F1 micro
	args:
	average: micro
	- type: f1
	name: F1 weighted
	args:
	average: weighted
	- type: precision
	name: Precision macro
	args:
	average: macro
	- type: precision
	name: Precision micro
	args:
	average: micro
	- type: precision
	name: Precision weighted
	args:
	average: weighted
	- type: recall
	name: Recall macro
	args:
	average: macro
	- type: recall
	name: Recall micro
	args:
	average: micro
	- type: recall
	name: Recall weighted
	args:
	average: weighted
	configs:
	- Ade_corpus_v2_classification
	- Ade_corpus_v2_drug_ade_relation
	- Ade_corpus_v2_drug_dosage_relation
	---

CI failures are only related to dataset cards missing some content.