Bare `except:` catches BaseException including KeyboardInterrupt and
SystemExit. Replaced 5 instances with `except Exception:`.

* Enhance SGDP optimizer with caution parameter

Added 'caution' parameter to SGDP optimizer for enhanced functionality.

* Add csgdp optimizer to the optimizer factory

* Update the link to the permanent URL of the accepted paper

Fix reference link for Spherical Cautious Optimizers

* Add tests for csgdp and csgdw optimizers

* Fix：Clone buffer for Nesterov momentum update

Clone the buffer before using it for the update.

…pace to avoid failures on timm train checkpoints

…ntion dispatch

The global attention path in MaskUnitAttention.forward() used a 5D tensor
reshape with num_windows=1 as a shortcut. This caused PyTorch SDPA to
silently fall back from efficient backends (FlashAttention, Memory-Efficient,
CuDNN) to the O(N^2) math backend, as all efficient kernels require 4D
contiguous tensors.

At high resolutions (e.g. 2048x2048 -> 16384 tokens), the math backend
materializes the full N*N attention matrix, causing catastrophic VRAM usage
and OOM on consumer GPUs.

Changes:
- Branch forward() into windowed (5D, unchanged) and global (4D) paths
- Global path reshapes directly to [B, N, 3, heads, head_dim] -> 4D QKV
- Adjust q_stride pooling dim from amax(dim=3) to amax(dim=2) for global
- Add .contiguous() on q, k, v to guarantee FlashAttention compatibility
- Split output transpose: transpose(1,3) for windowed, transpose(1,2) for global

…lated blocks