[BUG] Graspologic does not work on Python 3.10

[Dax Pryce]

Expected Behavior

We should be able to run our entire test suite using Python 3.10

Actual Behavior

graspologic/embed/n2v.py:11: in <module>
    from gensim.models import Word2Vec
/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/gensim/__init__.py:11: in <module>
    from gensim import parsing, corpora, matutils, interfaces, models, similarities, utils  # noqa:F401
/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/gensim/corpora/__init__.py:6: in <module>
    from .indexedcorpus import IndexedCorpus  # noqa:F401 must appear before the other classes
/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/gensim/corpora/indexedcorpus.py:14: in <module>
    from gensim import interfaces, utils
/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/gensim/interfaces.py:19: in <module>
    from gensim import utils, matutils
/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/gensim/matutils.py:10[24](https://github.com/microsoft/graspologic/runs/5147181977?check_suite_focus=true#step:5:24): in <module>
    from gensim._matutils import logsumexp, mean_absolute_difference, dirichlet_expectation
gensim/_matutils.pyx:1: in init gensim._matutils
    ???
E   ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

From: https://github.com/microsoft/graspologic/runs/5147180666?check_suite_focus=true#step:5:1

[Ben Pedigo]

crap, this is only on 3.10? it looks just like the one we had last week which makes me scared

[Dax Pryce]

Yup, and it concerns (and vexes) me too. I'm starting to lose faith in gensim being able to provide a word2vec algorithm that we can use over the long term. I really, really would rather not pull in another giant library like stellargraph though.