feat(pt): add AdaMuon optimizer (#5130)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Added AdaMuon optimizer for PyTorch training with mixed 2D/1D
parameter handling, batched orthogonalized updates, momentum and
per-parameter state.

* **Configuration**
* New optimizer options: momentum, weight decay, Adam betas, lr
adjustment modes and tuning coefficients; AdaMuon selectable as an
optimizer.

* **Tests**
* Added comprehensive tests for orthogonalization, optimizer updates,
bucketing, lr-adjust modes, weight decay, closure behavior, and state
save/load.

<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: OutisLi

deepmd/pt/optimizer/__init__.py

@@ -1,9 +1,12 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
+from .adamuon import (
+    AdaMuonOptimizer,
+)
 from .KFWrapper import (
     KFOptimizerWrapper,
 )
 from .LKF import (
     LKFOptimizer,
 )
 
-__all__ = ["KFOptimizerWrapper", "LKFOptimizer"]
+__all__ = ["AdaMuonOptimizer", "KFOptimizerWrapper", "LKFOptimizer"]