Optimize particle_waves_v6 hot-path allocations

[mochell]

Summary

This PR isolates allocation/performance-focused changes in particle_waves_v6.jl into a standalone review.

Changes

• Avoid per-call closure allocation for propagation correction fallback.
• Reuse αₚ intermediate instead of recomputing.
• Replace scalar broadcast expressions with scalar arithmetic.
• Remove temporary vector/slice allocations in propagation update by writing dz[4] and dz[5] directly.

Motivation

Keeps optimization-specific changes separate from the ongoing forcing-structure refactor branch, so each can be reviewed and merged independently.