JIT / TorchScript should support for sum(List[torch.Tensor]) like non-JIT PyTorch already does

[ProGamerGov]

🚀 The feature, motivation and pitch

The following code currently fails due to JIT not supporting the summarization of tensor lists:

from typing import List
import torch

def test_func(x: List[torch.Tensor]) -> torch.Tensor:
    return sum(x)

test_input = [torch.randn(1,2,4,4) for i in range(5)]

out_test = test_func(test_input) # Works

jit_test = torch.jit.script(test_func) # Fails

out_jit = jit_test(test_input)

It would be nice if JIT would work here for use cases like these.

Alternatives

No response

Additional context

This feature already works in normal PyTorch and it's simple enough that many would probably expect it to work in JIT.

[vadimkantorov]

regular sum also seems not very nice in terms of computational complexity (unless JIT can optimize it). A related feature request: #27522