No .exp() method in Torch Backend

[WeiguoMa]

Issue Description

It seems that there is no exp() method in torch backend, both in tensorcircuit-ng.backend and tensornetwork-ng. This might be caused by historical factors. Generally, I do think this function should be added in tensornetwork-ng.

How to Reproduce

import tensorcircuit as tc
tc.set_backend("pytorch")
c = tc.Circuit(2)
c.u(0)

Error Output

File /opt/miniconda3/envs/weiguoQ/lib/python3.12/site-packages/tensornetwork/backends/abstract_backend.py:833, in AbstractBackend.exp(self, tensor)
    825 def exp(self, tensor: Tensor) -> Tensor:
    826     """
    827     Return elementwise exp of `tensor`.
    828     Args:
   (...)    831       Tensor
    832     """
--> 833     raise NotImplementedError(
    834         "Backend '{}' has not implemented `exp`.".format(self.name)
    835     )

NotImplementedError: Backend 'pytorch' has not implemented `exp`.

Environment Context

OS info: macOS-26.0-arm64-arm-64bit
Python version: 3.10.18
Numpy version: 1.26.4
Scipy version: 1.15.2
Pandas version: 2.3.1
TensorNetwork version: 0.5.1
Cotengra version: 0.7.5
TensorFlow version: 2.15.0
TensorFlow GPU: []
Jax version: 0.6.2
Jax installation doesn't support GPU
JaxLib version: 0.6.2
PyTorch version: 2.2.2
PyTorch GPU support: False
PyTorch GPUs: []
Cupy is not installed
Qiskit version: 0.46.3
Cirq version: 1.4.1
TensorCircuit version 1.4.0

[refraction-ray]

Fixed, also add impl for torch expm method, which is absent five years ago in pytorch

[WeiguoMa]

Haha, I saw that comment you left in torch_backend