Support asymmetric padding for "same" padding with even kernel size

[dfsfdfse]

Describe the bug
PaddingConfig::Same doesn't return the same input shape when kernel_size = [2, 2]

Additional context

use burn::backend::Wgpu;
use burn::backend::wgpu::WgpuDevice;
use burn::tensor::Distribution;

let device = WgpuDevice::BestAvailable;
let data: Tensor<Wgpu, 4> = Tensor::random([1, 3, 16, 16], Distribution::Default, &device);
let pool = MaxPool2dConfig::new([2, 2]).with_padding(PaddingConfig2d::Same).init();
let v = pool.forward(data);
println!("{:?}", v.shape());

[laggui]

Thanks for flagging!

For even kernel sizes the "same" padding scheme actually requires asymmetric padding in order to produce the same output shape.

We only support symmetric padding right now, might be a good idea to add a check in the meantime 😅

If you try a kernel size of 2, 4, 6 or any other even number with same padding I believe you will observe the same behavior.