[TLE] Add optional offset parameter to tle.remote op

[lzllx123]

Overview

Add an optional scalar i64 offset parameter to the tle.remote op on the triton v3.6.x branch, allowing device-space remote pointers to access peer memory at a specified byte offset.

Changes

1. Add an optional scalar i64 offset parameter to tle.remote.
2. Restrict the offset to scalar i64 values only.
3. Use TT_IntLike for the RemotePointersOp offset operand to align with the verifier.
4. Enforce the offset requirement for device-space remote pointers and reject invalid cases.
5. Fix optional operand validation in the RemotePointersOp verifier.
6. Update test_tle_distributed_d2d.py to demonstrate the new offset parameter in tle.remote.