Sporadic "Initialization failed" error.

[nbarlowATI]

As reported in AIrsenal [https://github.com/alan-turing-institute/AIrsenal/issues/303], with bpl 0.1.0, we sometimes see the following when fitting the team model:

raceback (most recent call last):
  File "/Users/nbarlow/anaconda3/envs/airsenv2/lib/python3.7/multiprocessing/pool.py", line 121, in worker
    result = (True, func(*args, **kwds))
  File "/Users/nbarlow/anaconda3/envs/airsenv2/lib/python3.7/multiprocessing/pool.py", line 44, in mapstar
    return list(map(*args))
  File "stanfit4anon_model_2ed431ea7c899defb6dfb9552dda195d_1299729220150895031.pyx", line 373, in stanfit4anon_model_2ed431ea7c899defb6dfb9552dda195d_1299729220150895031._call_sampler_star
  File "stanfit4anon_model_2ed431ea7c899defb6dfb9552dda195d_1299729220150895031.pyx", line 406, in stanfit4anon_model_2ed431ea7c899defb6dfb9552dda195d_1299729220150895031._call_sampler
RuntimeError: Initialization failed.

To get the data to reproduce this:

# clone AIrsenal and checkout develop branch
git clone https://github.com/alan-turing-institute/AIrsenal
cd AIrsenal
git checkout develop
pip install .
python
>>> from airsenal.conftest import test_past_data_session_scope
>>> from airsenal.framework.bpl_interface import *
>>> with test_past_data_session_scope() as testsession:
>>>    results_df = get_result_df("1819", 10, testsession)
>>>    ratings_df = get_ratings_df("1819", testsession)
>>> # instantiate and fit the bpl model
>>> import bpl
>>> model_team = bpl.BPLModel(results_df, ratings_df)
>>> model_team.fit()