Add Normal/MultivariateNormal PDFs

[jackaraz]

This PR extends the PDF list with normal and multivariate normal distributions:

spey.AvailableBackends()
# ['pyhf',
#  'pyhf.simplify',
#  'pyhf.uncorrelated_background',
#  'default_pdf.correlated_background',
#  'default_pdf.effective_sigma',
#  'default_pdf.multivariate_normal',
#  'default_pdf.normal',
#  'default_pdf.poisson',
#  'default_pdf.third_moment_expansion',
#  'default_pdf.uncorrelated_background']


pdf_wrapper = spey.get_backend("default_pdf.multivariate_normal")
statistical_model1 = pdf_wrapper(
    signal_yields=[12.0, 15.0],
    background_yields=[50.0,48.0],
    data=[36, 33],
    covariance_matrix=[[144.0,13.0], [25.0, 256.0]],
)

pdf_wrapper = spey.get_backend("default_pdf.normal")
statistical_model2 = pdf_wrapper(
    signal_yields=[12.0, 15.0],
    background_yields=[50.0,48.0],
    data=[36, 33],
    absolute_uncertainties=[20.0,10.0],
)

print(statistical_model2.chi2())
# 7.949999999999935

print(statistical_model1.exclusion_confidence_level(), statistical_model2.exclusion_confidence_level())
# ([0.9591145211259625], [0.9890014596358632])

[jackaraz]

Note: Autograd does not work with numpy v2.0.0. Thus numpy version has been limited to be less than 2.