Add laplacian operator

[mavenlin]

Immediate solution

Laplace operator is important for computing the kinetic energy, currently we have to use two nabla operators together with the trace. For example

1. in the transpose rule of nabla,
   https://github.com/sail-sg/autofd/blob/main/autofd/operators/operators.py#L773

2. in mgga functional of jax_xc,
   https://github.com/sail-sg/jax_xc/blob/main/gen_repo/experimental.jinja#L66

While we can do better with the jet rules.

In the long term

In the long term, I believe we need to have a jaxpr level graph rewriting tool, we can then introduce algebraic relationships that works more generally for any nested nabla operations.

[mavenlin]

According to https://github.com/y1xiaoc/fwdlap/, jet is not the most efficient option for laplace calculation, a more efficient way would be to reduce at every op.

We can use fwdlap if y1xiaoc/fwdlap#2 is resolved.