resultAnalysis crashes when prediction_samples is specified

[Francesco-Baroni]

Problem:
When the resultAnalysis function is called with the prediction_samples argument, it crashes.
A solution is to also set the batch_size argument to a value smaller than the value of the prediction_samples argument. However, changing the batch_size in the range 0 < batch_size <= prediction_samples also alters the result of the analysis.

Code examples:
model.resultAnalysis(dataset='data_test', name='Model', prediction_samples=100) This crashes.
model.resultAnalysis(dataset='data_test', name='Model', batch_size=128, prediction_samples=100) This crashes.
model.resultAnalysis(dataset='data_test', name='Model', batch_size=100, prediction_samples=100) This gives some analysis results.
model.resultAnalysis(dataset='data_test', name='Model', batch_size=10, prediction_samples=100) This gives different analysis results.

The dataset used has 60000 samples.

[MisterMandarino]

import os
from nnodely import *

x = Input('x')
F = Input('F')
out_rel = Fir(x.tw(1)) + Fir(F.last())
x.closedLoop(out_rel)
x_z_est = Output('x_z_est', out_rel)

mass_spring_damper = Modely(seed=42, visualizer=TextVisualizer())
mass_spring_damper.addModel('x_z_est', x_z_est)
mass_spring_damper.addMinimize('next-pos', x.z(-1), x_z_est, 'mse')
mass_spring_damper.neuralizeModel(0.2)

data_struct = ['time','x','dx','F']
data_folder = os.path.join(os.path.dirname(os.path.realpath(file)),'dataset','data')
mass_spring_damper.loadData(name='mass_spring_dataset', source=data_folder, format=data_struct, delimiter=';')

mass_spring_damper.resultAnalysis('mass_spring_dataset', prediction_samples=10, batch_size=5) # gives an error
mass_spring_damper.resultAnalysis('mass_spring_dataset', prediction_samples=10, batch_size=15) # gives a different error