[Feature][QDP] Add MPI-ready distributed amplitude execution scaffolding

[viiccwen]

Related Issues

• None yet.

Changes

• add distributed amplitude planning, layout, and state scaffolding for multi-GPU execution
• introduce a distributed execution context and collective seam that are MPI-ready without requiring an MPI runtime yet
• fix placement-to-device binding so reordered placements materialize buffers on the correct CUDA devices
• add a distributed q34 probe and distributed runtime / planner / topology / communicator test coverage
• update CUDA arch targeting so the probe builds on the GPUs available on this host

Why

• establish a QDP-native distributed amplitude foundation that can scale beyond a single GPU
• keep the execution model extensible toward MPI-backed collectives without hard-coding Lightning-specific assumptions
• verify that the current implementation can successfully materialize and encode a 34-qubit float32 distributed state on this host

How

• keep planning, layout, runtime, and collective responsibilities separate
• drive shard materialization from placement device IDs rather than raw mesh iteration order
• package mesh plus collectives into a single distributed execution context
• validate the path with distributed tests and an executable q34 probe

Checklist

• Feature
• Tests added or updated
• No Python surface changes
• Scope kept to QDP distributed amplitude groundwork
• MPI multi-rank launcher path implemented

[viiccwen]

Closing this PR because the changes have been moved onto PR #1296 as intended.