Consolidate lang usage in pke

[ygorg]

• Fixed: Trying to load non available stopwords raises TypeError
• Fixed: Some langcodes in pke.lang.langcodes were not accurate (Lang Code "pt" for portuguese language #193, Fix for German and Polish stopwords #215, update lang.py to be usable with spanish #216, Fix some ISO-2 langcodes #219)
• Clarified what languages are supported in pke (添加"zh": "chinese" #191), only the one that have a nltk.stem.SnowballStemmer and a spacy model.
• Added exception to prevent using a keyword extractor that wasn't properly initialized.

There are 3 cases when using pke regarding to language used:

• spacy model and stemmer available

python -m spacy download fr_core_news_sm
python << BEGIN
import pke
e = pke.unsupervised.PositionRank()
e.load_document("Ga bu zo meu", language='fr')
BEGIN

• spacy model but no stemmer available : falls back to porter stemmer, does not use stopwords

python -m spacy download pl_core_news_sm
python << BEGIN
import pke
e = pke.unsupervised.PositionRank()
e.load_document("Ga bu zo meu", language='pl')
# WARNING:root:No stoplist available in pke for 'pl' language.
# WARNING:root:No stemmer available in pke for 'pl' language -> falling back to porter stemmer.
BEGIN

• no spacy model available: preprocessing needs to happen outside of pke

python << BEGIN
import pke
e = pke.unsupervised.PositionRank()
doc = [[('Ga', 'DET'), ('bu', 'ADJ'), ('zo', 'NOUN'), ('meu', 'VERB')]]
e.load_document(doc, language='ug')
# WARNING:root:No stoplist available in pke for 'ug' language.
# WARNING:root:No stemmer available in pke for 'ug' language -> falling back to porter stemmer.
BEGIN