Better support for bounded/constrained distributions

[danielward27]

I am happy leaving it to the user to enforce constraints they require, rather than having explicitly constrained distributions, however:

• We need a Log/Exp bijection
• We should ensure that Distribution.log_prob calculations involving arctanh and log return -jnp.inf where appropriate (for samples out of support), rather than jnp.nan.
• We should have an example of learning a distribution with bounded support