p-values on env components about s-jsdm

[MaximilianPi]

It is done, but there are still some issues:

The computation of the hessian is exhaustive: for each species and each env, we have to do a full run which results in a higher runtime for the hessian than for the actual model fit. Thats why, I introduced a "se" argument in the sjSDM function which turns on/off the standard error computation with default = FALSE.

[MaximilianPi]

Yeah, and calculating hessian per species is more efficient for large communities (for 1000 species, we would have to calculate 1000^2 hessian...).

I added also an 'getSe' function to calculate the standard errors post hoc bbbdfb4

[florianhartig]

So, this could be closed, right?