Improve naming

[kengelbrecht]

• NLUdataset.to_json/from_json are not compatible -> change interface so that what has been written with to_json() can be loaded with from_json()

Optimierung von Namen

• rasa -> rasa2 (because the outdated rasa version should not be named like the default one)
• NLUdataset -> NluDataset
• NLUdataset.from_joined() -> NLUdataset.concat() (bzw. äquivalent zu Pandas benennen)
• Make classifier names more consistent.
  - TfidfIntentClassifier -> OK
  - SpacyClassifier -> Spacy
  - TelekomModel -> CharNgramIntentClassifier
  - Rasa2/3 -> OK
  - Watson -> WatsonAssistant
  - LUIS -> Luis
  - FastText -> OK
• Clean up vendor file names
  - telekom -> char_ngram_intent_classifier
  - watson -> watson_assistant
• Change datasets.py -> nlu_dataset.py to avoid name conflicts with huggingface datasets and be more consistent in nameing (file called like class) and also change vendors.py -> vendor.py for consistency with nlu_dataset

Reasoning behind the vendor name changes:

• Should be short
• Should not make claims that may not hold in the future (e.g. SpacyClassifier might in fact be extended with entity recognition)
• TelekomModel was renamed to avoid branding issues and avoid someone believing Telekom sells this.
• Watson renamed to WatsonAssistant as Watson has also other NLU products, so was unclear which is used before
• LUIS to Luis so Acronyms are written in the same way everywhere
• Added "IntentClassifier" for our own implementations to add clarity (these will not be extended with entity recognition)

[kengelbrecht]

Created branch feature/better_names

• Rasa now Rasa2 (also changed file name to rasa2)
• NLUdataset now NluDataset

[kengelbrecht]

Added function nlubridge.concat(dataset_list) analoguous to pandas.concat().
Added NluDataset method joined = NluDataset.join(other_dataset) analoguous to pandas.DataFrame.join().
Added DeprecationWarning for NluDataset method NluDataset.from_joined; should use one of the new methods (most likely concat() in this case)

[kengelbrecht]

to_json/from_json already works correctly. Added a test to assert from_json works with the string returned by to_json(path=None).

[kengelbrecht]

Changes as outlined in the description merged to develop with commit 9e93bc8