Allow transformation of weak vars

[jobrachem]

So far, we did not allow transformation of weak vars in Var.transform. I have a usecase for that right now, and I don't think anything really speaks against allowing it. This PR makes two small changes that enable the functionality.

If you transform a weak var, it gets inserted in between the old value node and the transformed variable:

import liesel.model as lsl
import jax.numpy as jnp
import tensorflow_probability.substrates.jax.bijectors as tfb

tau = lsl.Var.new_param(10.0, name="tau")
tau_sqrt = lsl.Var.new_calc(jnp.sqrt, tau, name="tau_sqrt")
log_tau_sqrt = tau_sqrt.transform(tfb.Exp())
model = lsl.Model([tau_sqrt])
lsl.plot_vars(model)