[MPS] Add lerp implementation (#105470)
lerp.Scalar fits very well into binary op template
Add a very naive implementation for `lerp.Tensor` as `add_out(self, weights.mul(end.sub(self)))`
Enable `lerp` testing in `test_mps`
Fixes https://github.com/pytorch/pytorch/issues/105382
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105470
Approved by: https://github.com/albanD
