Bug in units of return value of delta_sigma

[aphearin]

At a minium, as pointed out by @alexieleauthaud, the units of the return value of the delta_sigma function are not correct. One clear contribution to the problem is that the mean matter density is being set to Nptcl/Vbox, whereas this density must be multiplied by the particle mass to have the correct dimensions. However, resolving this bug should instead implement the mean matter density to be determined by the cosmology of the box. See the empirical_models.phase_space_models.profile_models.profile_helpers.density_threshold function for an example of how this is done.

CC @duncandc - I'm self-assigning this one, but I'll ping you to look at this with me next week.

[aphearin]

@alexieleauthaud - this bug is now resolved since I have just merged in #537. The delta_sigma function in the master branch is now ready to use.