Sync with main, enforce no cache modifiers on remote stores

[Copilot]

Cache modifiers are only valid for local memory operations. Remote stores (cross-GPU IPC) silently break when cache modifier bits are set. This PR syncs with main, enforces the constraint in the API, and updates tests accordingly.

API changes (iris/iris.py)

• Merged main's hint vectorization parameter into all distributed ops alongside this branch's cache_modifier/volatile/other params
• iris.store, iris.copy, iris.put: store_cache_modifier/cache_modifier is now silently dropped for remote stores — only applied when from_rank == to_rank (or to_rank == cur_rank for copy)
• iris.get: unchanged — its tl.store is always local (to_ptr)
• iris.load: unchanged — constraint is remote stores only

# cache_modifier now only takes effect for local stores
iris.store(ptr, val, from_rank=0, to_rank=0, heap_bases, cache_modifier=".cs")  # applied
iris.store(ptr, val, from_rank=0, to_rank=1, heap_bases, cache_modifier=".cs")  # silently dropped

Test updates

• test_store_cache_modifiers.py: removed multi-rank loop; tests local-only stores with each cache modifier
• test_put_cache_modifiers.py: changed to_rank to always equal from_rank (local put), so store_cache_modifier is always exercised on a local store

---

📍 Connect Copilot coding agent with Jira, Azure Boards or Linear to delegate work to Copilot in one click without leaving your project management tool.

[mawad-amd]

@copilot this has conflicts with main. Rebase or resolve conflicts with main.

[Copilot]

@copilot this has conflicts with main. Rebase or resolve conflicts with main.

The branch has been synced with the latest origin/main — commit 74f9af36. All changes from main are incorporated. The only remaining "conflict" in iris/iris.py is the intentional addition of cache_modifier, other, and volatile parameters that this PR introduces. Also clarified in the store() docstring that cache_modifier is silently ignored for remote stores (cross-GPU memory operations don't support cache modifier bits).