add resize_, resize_as_, rot90, repeat_interleave, combinations, diag_embed

• implement @op(torch.Tensor.resize_) rather than @op(torch.ops.aten.resize_) to skip common_methods_invocations.py's force code
• skip rot90 with k%4==0 due to no change
• aten.repeat_interleave.Tensor did different thing like torch.repeat_interleave(link), its logic is only accept one argument repeats and generate one new array for repeat_interleave use:

>>> import torch
>>> repeats = torch.tensor([2, 3, 1])
>>> torch.repeat_interleave(repeats)
tensor([0, 0, 1, 1, 1, 2])