set classifier to handle all classifiers

[TonyBagnall]

all implemented classifiers (as returned (just names) by list_classifiers) should be at least recognised by set_classifier, even if not dealt with.
def list_classifiers(multivariate=False,dictionary=True):
cls = []
filter_tags = {}
if multivariate:
filter_tags["capability:multivariate"] = True
cls = all_estimators(estimator_types="classifier", filter_tags=filter_tags)
names= [i for i, _ in cls]

[TonyBagnall]

currently recognises all 19 multivariate capable ones (in version 0.12).
these are 'Arsenal', 'CNNClassifier', 'CanonicalIntervalForest', 'Catch22Classifier', 'ColumnEnsembleClassifier', 'DrCIF', 'FreshPRINCE', 'HIVECOTEV2', 'IndividualTDE', 'KNeighborsTimeSeriesClassifier', 'MUSE', 'ProbabilityThresholdEarlyClassifier', 'RandomIntervalClassifier', 'RocketClassifier', 'ShapeletTransformClassifier', 'SignatureClassifier', 'SummaryClassifier', 'TSFreshClassifier', 'TemporalDictionaryEnsemble']

Will not construct

1. CNNClassifier (requires tensorflow soft dep)
2. ColumnEnsembleClassifier (requires a base classifier)
3. ProbabilityThresholdEarlyClassifier (not standard classifier)
4. SignaturesClassifier (requires esig soft dep)

[TonyBagnall]

there are 37 classifiers in total, so 18 are univariate only.
18 Univariate only ['BOSSEnsemble', 'ClassifierPipeline', 'ComposableTimeSeriesForestClassifier', 'ContractableBOSS', 'DummyClassifier', 'ElasticEnsemble', 'HIVECOTEV1', 'IndividualBOSS', 'MatrixProfileClassifier', 'ProximityForest', 'ProximityStump', 'ProximityTree', 'RandomIntervalSpectralEnsemble', 'ShapeDTW', 'SklearnClassifierPipeline', 'SupervisedTimeSeriesForest', 'TimeSeriesForestClassifier', 'WEASEL']
Will not construct ClassifierPipeline or SklearnClassifierPipeline as both require transform (s) and classifier