Feature request to implement ~.cdf() method in MultivariateNormalTriL class

System information

OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Linux Ubuntu 21.04
TensorFlow installed from (source or binary): binary
TensorFlow versions:

$ python -c "import tensorflow as tf; import tensorflow_probability as tfp; print(tf.version.GIT_VERSION, tf.version.VERSION, tfp.__version__)"
v2.5.0-rc3-213-ga4dfb8d1a71 2.5.0 0.13.0

Issue

I am trying to get the cumulative distribution function for a bivariate normal distribution, here is the code:

import tensorflow as tf
import tensorflow_probability as tfp
tfd = tfp.distributions
tf.compat.v1.enable_eager_execution()

mvn = tfd.MultivariateNormalTriL(loc=[0,0], scale_tril=tf.linalg.cholesky([[1,0.5],[0.5,1]]))
mvn.cdf([0,0])

And here is the error that I get:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/distributions/distribution.py", line 1438, in cdf
    return self._call_cdf(value, name, **kwargs)
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/distributions/distribution.py", line 1414, in _call_cdf
    return self._cdf(value, **kwargs)
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/distributions/transformed_distribution.py", line 407, in _cdf
    self.bijector._internal_is_increasing(**bijector_kwargs),  # pylint: disable=protected-access
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/bijector.py", line 1171, in _internal_is_increasing
    return self._call_is_increasing(name, **kwargs)
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/bijector.py", line 1155, in _call_is_increasing
    return tf.identity(self._is_increasing(**kwargs))
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/chain.py", line 147, in _is_increasing
    is_increasing, b._internal_is_increasing(**kwargs.get(b.name, {})))  # pylint: disable=protected-access
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/bijector.py", line 1171, in _internal_is_increasing
    return self._call_is_increasing(name, **kwargs)
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/bijector.py", line 1155, in _call_is_increasing
    return tf.identity(self._is_increasing(**kwargs))
  File "/home/alexlembcke/.local/lib/python3.9/site-packages/tensorflow_probability/python/bijectors/bijector.py", line 1150, in _is_increasing
    raise NotImplementedError('`_is_increasing` not implemented.')
NotImplementedError: `_is_increasing` not implemented.

If I adapt the code to use scipy, like so:

from scipy.stats import multivariate_normal

mvn = multivariate_normal(mean=[0,0], cov=[[1,0.5],[0.5,1]])
mvn.cdf([0, 0])

Then everything works fine. Is this a bug or am I doing something wrong in my tensorflow code?

Start at the MultivariateNormalTriL cdf entry point and reproduce the supplied TensorFlow example, comparing its expected behavior with the SciPy example. Done means the bivariate normal CDF request works and the behavior is covered by regression tests.