Support scipy sparse matrices

[aguschin]

Support https://docs.scipy.org/doc//scipy/reference/sparse.html matrices/arrays as data formats for MLEM

Required to add MLEM to https://github.com/iterative/example-get-started

To reproduce, run dvc repro for the branch from aguschin/example-get-started#1

Traceback (most recent call last):
  File "/Users/aguschin/Git/iterative/example-get-started-gto/src/train.py", line 47, in <module>
    save(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/telemetry.py", line 50, in inner
    return f(*args, **kwargs)
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/metadata.py", line 120, in save
    meta = get_object_metadata(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/metadata.py", line 53, in get_object_metadata
    return MlemModel.from_obj(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/objects.py", line 758, in from_obj
    sample_data, methods_sample_data = mlem_model._create_preprocessors(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/objects.py", line 687, in _create_preprocessors
    pre, pre_method = self._create_processor(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/objects.py", line 655, in _create_processor
    model_type = ModelAnalyzer.analyze(obj, sample_data=sample_data).bind(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/model.py", line 358, in analyze
    return super().analyze(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/hooks.py", line 107, in analyze
    return cls.find_hook(obj).process(obj, **kwargs)
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/contrib/callable.py", line 213, in process
    s = Signature.from_method(
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/model.py", line 212, in from_method
    returns = DataAnalyzer.analyze(result)
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/hooks.py", line 107, in analyze
    return cls.find_hook(obj).process(obj, **kwargs)
  File "/Users/aguschin/.local/share/virtualenvs/gto-mP9zWfJO/lib/python3.9/site-packages/mlem/core/hooks.py", line 135, in find_hook
    raise HookNotFound(
mlem.core.errors.HookNotFound: No suitable DataHook for object of type "csr_matrix". Registered hooks: [<class 'mlem.core.data_type.PrimitiveType'>, <class 'mlem.core.data_type.OrderedCollectionHook'>, <class 'mlem.core.data_type.DictTypeHook'>, <class 'mlem.core.data_type.BinaryDataType'>, <class 'mlem.contrib.numpy.NumpyNumberType'>, <class 'mlem.contrib.numpy.NumpyNdarrayType'>]

under #423

[aguschin]

Ok, so this one was actually fixed in #538. MLEM still doesn't support other formats except for CSR, but that's fine for now. We can add them later when needed.