Added operations for the spectrum class

[fnemina]

Added common arithmetic operations as well as numpy functions to be used
in the SPECTRUM class. To do this we defined the array_ufunc method
as recommended here

https://numpy.org/doc/stable/reference/arrays.classes.html

as well as changed the kind to numpy.lib.mixins.NDArrayOperatorsMixin to
allow for the normal operations to be handled by numpy.

An SPECTRUM class is returned with the same metadata, a new name and
a 'TRANS_TYPE' measure_type.

Also it was defined how the array method should be handled.

This should make it easier to operate with SPECTRUM data, for example

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from specdal import *

measurement1 = pd.Series([1, 2, 3, 4], index=pd.Index([1, 2, 3, 4], name='wavelength'),
                                name='pct_reflect')
measurement2 = pd.Series([2, 4, 6, 8], index=pd.Index([1, 2, 3, 4], name='wavelength'),
                                name='pct_reflect')
s1 = Spectrum(name='s1', measurement=measurement1)
s2 = Spectrum(name='s2', measurement=measurement2)

s = (s1+s2)/2

s1.plot(label="s1")
s2.plot(label="s2")
s.plot(label="average")
plt.legend()
plt.show()

returns in s an SPECTRUM class object with the average of both spectra. Also more
complicated functions as np.sin(s) can be calculated.

It was also incorporated the __array__ method, so that functions as np.mean(s)
returns the means over all wavelengths.