Bug with zero values in computing DENSITY bins

[yoid2000]

There is a bug whereby sdnist crashes if there are zero values in the synthetic data for DENSITY.

The problem is that, although you bottom code DENSITY to zero here:

SDNist/sdnist/report/dataset/binning.py

Line 102 in 8430da9

d.loc[d['DENSITY'] < 0, 'DENSITY'] = float(0)

The left-most bin boundary of 0.0 is not inclusive at the pd.cut() function:

SDNist/sdnist/report/dataset/binning.py

Lines 105 to 109 in 8430da9

	if update:
	d['DENSITY'] = pd.cut(d['DENSITY'], bins=bins, labels=labels)
	return d
	else:
	d['binned_density'] = pd.cut(d['DENSITY'], bins=bins, labels=labels)

As a result, any 0.0 DENSITY values in the synthetic data are set to NaN, and then later in transform.py when the values are set to numeric:

SDNist/sdnist/report/dataset/transform.py

Line 32 in 8430da9

data[c] = pd.to_numeric(data[c]).astype(int)

The NaN values cause a crash.

The solution is to make the left-most bin boundary inclusive:

    if update:
        d['DENSITY'] = pd.cut(d['DENSITY'], bins=bins, labels=labels, include_lowest=True)
        return d
    else:
        d['binned_density'] = pd.cut(d['DENSITY'], bins=bins, labels=labels, include_lowest=True)