Multidimensional Example

It would be nice to expand the README to show how to use autobound with vector-valued inputs and interpret the output.
From looking into the code, I thought that

```python
import autobound.jax as ab
import jax.numpy as jnp
f = lambda x: 1.5*jnp.exp(3.0*x[1]*x[0]) - 25.0*x[0]**2
trust_region = (jnp.array([0,0]),jnp.array([1,1]))
x0 = jnp.array([.5,.5])
bounds = ab.taylor_bounds(f,max_degree=2)(x0,trust_region)
```
should work, but that runs into issues with `dynamic_slice` not being implemented.
The following worked:

```python
A= jnp.array([0,1])
B= jnp.array([1,0])
f = lambda x: 1.5*jnp.exp(3.0*jnp.dot(A,x)*jnp.dot(B,x)) - 25*jnp.dot(B,x)**2
```

Perhaps there is an intended, easier way I am missing?