Add warning not to intialize walkers in same starting place

[sevyharris]

A PR to add an extra warning against user error if they initialize all the walkers in the same place. Based on this issue.
#38

Here's the code I was using to test a correct vs incorrect initialization:

import numpy as np
from sklearn.datasets import make_spd_matrix
import matplotlib.pyplot as plt
import zeus


ndim = 10
nwalkers = 30
nsteps= 5000

C = make_spd_matrix(ndim)

icov = np.linalg.inv(C)

mu = np.random.rand(ndim)

def log_prob(x, mu, icov):
    return -0.5 * np.dot(np.dot((x-mu).T,icov),(x-mu))

# correct initialization
start = np.random.randn(nwalkers, ndim)


# incorrect initialization
start = np.random.randn(1, ndim)
start = np.tile(start, (nwalkers, 1))

print(start)


sampler = zeus.EnsembleSampler(nwalkers, ndim, log_prob, args=[mu, icov])
sampler.run_mcmc(start, nsteps)


print('Samples: ', sampler.samples.flatten(discard = 0).shape)