Variance gamma process sample depends on the step number

[gthewhite]

Hello, I believe that _sample_variance_gamma_process generates realizations that depends on the number of the steps at fixed maximum time.
If I change the n parameter from 100 to 1000 I obtain that the excursion of the trajectories change from [-0.002 : 0.002] to [-0.0006 to 0.0006]. This gives a 2nd moment depending on the time step of the realization. It sounds bad.
The other parameters are:
VarianceGammaProcess(drift=0, variance=10., scale=0.002, t=1, rng=None)
Would it be possible to fix this?
Thanks