[math] Add vector filling overload for scalar distributions of TRandom

[ericcano]

[math] Add new array filling functions for distributions delivering scalars in TRandom

Changes or fixes:

Addition of TRandom::<Distribution>N() array filler functions.

>>> vd = np.zeros(100, dtype=np.float64)
>>> r = ROOT.TRandom2(123)
>>> r.GausN(100, vd, 0, 1.0)
>>> vd
array([-0.51452618,  0.11320471, -0.2552159 , -0.9487521 ,  1.25042753,
        -0.34611829,  2.3636373 ,  1.52963734,  0.95990059,  0.21563747,
        -0.48241846,  1.51473977, -1.19058241, -1.62088594,  0.57052152,
         0.15026036,  1.58667136,  0.36062987, -1.23310079,  1.20035556,
        -0.42508447,  0.13440585,  0.63766076,  0.59627148, -0.46659047,
        -1.57966641, -0.23080729, -0.05128992,  1.27449527, -0.3358339 ,
        -2.53090467, -1.54203658, -0.84014657,  0.24656162,  0.08736865,
         0.66275005, -1.47442608, -0.48096595,  0.72262188,  1.07080308,
         0.24917973, -0.4976157 ,  0.47648086, -0.89377689,  0.82566228,
        -1.26325485, -0.17795292,  0.74986627,  1.16974563,  1.41039828,
        -0.01727918, -0.08254512,  0.39879297,  0.73254844,  0.80943978,
         1.79774612,  1.80436324, -1.13506493, -1.18507172, -0.63729355,
         1.08810798,  0.40591267,  0.03376047,  0.49296303,  2.89776279,
        -1.77849263, -0.3847957 , -0.15786366, -0.21491428, -0.9693155 ,
         1.50609877,  0.6414143 , -0.54620441,  1.363164  , -0.63699385,
         1.77694564,  0.62645766,  0.46607116,  1.12608175,  1.67122243,
        -0.40426394,  0.50792749,  0.49501287,  0.05635704,  1.66376808,
         0.06030797, -1.79686209,  1.9065689 , -0.60956203,  0.32817868,
        -1.14337713, -0.020879  ,  1.19166155, -1.55478791, -0.07289054,
         0.77477693, -0.93212149, -1.65628565,  0.29406467,  0.48126101])

This pythonisation is validated in unit tests.

We could not overload the existing functions as default values would create ambiguous cases.

Checklist:

• tested changes locally
• updated the docs (if necessary)

[aaronj0]

Looks good! We can use something like:

 def _TRandom_Uniform(self, ...):

     import ROOT
     import numpy
 
    #  if user expects a vector:
         return self.UniformN(...)
    # if matches original Uniform implementation
        return self._Uniform(...)
    
   @pythonization('TRandom')
 def pythonize_trandom(klass):
     klass._Uniform = klass.Uniform
     klass.Uniform = _TRandom_Uniform

To automatically forward to the loop wrappers when the user expects an array of random numbers, or to the original generation function when the input signatures follow the traditional format