Minor speedup using numba-stats

[jeethu]

Before:

$ DISABLE_NUMBA_STATS=1 python -m timeit -s "sz=5000;import numpy as np;import pandas as pd;from numerai_tools.scoring import numerai_corr;t = pd.Series(np.random.choice([0, 0.25, 0.5, 0.75, 1.0], size=sz), name='t');df = pd.DataFrame({'predictions': np.random.rand(sz)})" "numerai_corr(df, t)"

50 loops, best of 5: 3.9 msec per loop

After:

$ python -m timeit -s "sz=5000;import numpy as np;import pandas as pd;from numerai_tools.scoring import numerai_corr;t = pd.Series(np.random.choice([0, 0.25, 0.5, 0.75, 1.0], size=sz), name='t');df = pd.DataFrame({'predictions': np.random.rand(sz)})" "numerai_corr(df, t)"

100 loops, best of 5: 3.75 msec per loop

About 3.85% faster for computing numerai_corr(). I haven't benchmarked other functions.