[TorchToLinalg] Add support for AtenReplicationPad1dOp and lower to linalg backend

[zahidwx]

Problem

• AtenReplicationPad1dOp is not available in torch dialect. There is no lowering to linalg backend.
  https://docs.pytorch.org/docs/stable/generated/torch.nn.ReplicationPad1d.html
• Currently AtenPadOp decomposition in torch dialect uses replication pad 2d for cases which require the usage of 1d replication pad. Update the decomposition by making use of AtenReplicationPad1dOp for 1d padding with "replicate" mode.

[zahidwx]

Related PR: #4217 merged to main.