RDMA/mthca: Use generic resize-CQ lock

Replace the open‑coded resize‑CQ lock with the standard core
implementation for better consistency and maintainability.

Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
3 files changed