feat(dpa4): multiple updates for dpa4 (#5734)

Since each commits relies on each other to pass the parity tests, so
this is a fusion pr.

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

* **New Features**
* Added `readout_layers` and native per-atom spin support across
SeZM/DPA4 descriptors, including new `spin` runtime input and spin-aware
scalar readout.
* Introduced a public `SpinEmbedding` component for the native spin
scheme.
* Added `"fndc"` tensor-layout support and expanded optional fused
Triton acceleration (including flash-attention-style aggregation) for
supported configurations.
* **Bug Fixes**
* Improved descriptor behavior for zero-block/empty-edge cases;
strengthened full-validation profile handling and relaxed spin label
requirements when allowed.
* Reduced evaluation logging verbosity and enabled AMP autocast during
inference when opted in.
* **Compatibility**
  * Model deserialization now recognizes `sezm_native_spin`.
  * Removed an older experimental CuTe rotation path.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: OutisLi

deepmd/dpmodel/descriptor/dpa4.py

@@ -41,6 +41,7 @@
     EnvironmentInitialEmbedding,
     GeometricInitialEmbedding,
     SeZMTypeEmbedding,
+    SpinEmbedding,
 )
 from .ffn import (
     EquivariantFFN,
@@ -158,6 +159,7 @@
     "ScalarRMSNorm",
     "SeZMInteractionBlock",
     "SeZMTypeEmbedding",
+    "SpinEmbedding",
     "SwiGLU",
     "WignerDCalculator",
     "apply_lora_to_sezm",

deepmd/dpmodel/descriptor/dpa4_nn/__init__.py

@@ -92,7 +92,8 @@ class GatedActivation(NativeOP):
         Whether to use bias in the gate linear layer.
     layout
         Tensor layout convention. ``"nfdc"`` means input shape (N, F, D, C);
-        ``"ndfc"`` means input shape (N, D, F, C).
+        ``"ndfc"`` means input shape (N, D, F, C); ``"fndc"`` means input shape
+        (F, N, D, C), the focus-major layout used by the SO(2) mixing stack.
     trainable
         Whether parameters are trainable.
     seed
@@ -125,8 +126,8 @@ def __init__(
         self.precision = precision
         self.mlp_bias = bool(mlp_bias)
         self.layout = str(layout).lower()
-        if self.layout not in {"nfdc", "ndfc"}:
-            raise ValueError("`layout` must be either 'nfdc' or 'ndfc'")
+        if self.layout not in {"nfdc", "ndfc", "fndc"}:
+            raise ValueError("`layout` must be one of 'nfdc', 'ndfc', or 'fndc'")
 
         self.activation_function = str(activation_function)
         self.scalar_act = get_activation_fn(activation_function)
@@ -170,7 +171,8 @@ def call(self, x: Any, gate: Any = None) -> Any:
         ----------
         x
             Value features. Shape is (N, F, D, C) when ``layout='nfdc'``,
-            or (N, D, F, C) when ``layout='ndfc'``.
+            (N, D, F, C) when ``layout='ndfc'``, or (F, N, D, C) when
+            ``layout='fndc'``.
         gate
             Optional gate features with the same layout as ``x``.
             When provided, enables GLU mode:
@@ -184,6 +186,10 @@ def call(self, x: Any, gate: Any = None) -> Any:
             Gated features with the same layout as ``x``.
         """
         xp = array_api_compat.array_namespace(x)
+        # ``ndfc`` carries the degree axis at position 1; ``nfdc`` and the
+        # focus-major ``fndc`` carry it at position 2. Every select/narrow/reshape
+        # below is expressed against this single degree axis, so the three layouts
+        # share one code path apart from the per-focus gate projection.
         degree_axis = 1 if self.layout == "ndfc" else 2
 
         scalar_idx = tuple(
@@ -211,14 +217,17 @@ def call(self, x: Any, gate: Any = None) -> Any:
             return x0
 
         input_dtype = gate_scalar_source.dtype
-        gating_scalars = xp.astype(
-            xp_sigmoid(
-                self.gate_linear(
-                    xp.astype(gate_scalar_source, get_xp_precision(xp, self.precision))
-                )
-            ),
-            input_dtype,
-        )
+        gate_src = xp.astype(gate_scalar_source, get_xp_precision(xp, self.precision))
+        if self.layout == "fndc":
+            # The scalar source is focus-major (F, N, C). ``FocusLinear`` mixes
+            # channels with the focus stream on axis 1, so present it in the shared
+            # (N, F, C) convention and restore the focus-major orientation.
+            gate_logits = xp.permute_dims(
+                self.gate_linear(xp.permute_dims(gate_src, (1, 0, 2))), (1, 0, 2)
+            )
+        else:
+            gate_logits = self.gate_linear(gate_src)
+        gating_scalars = xp.astype(xp_sigmoid(gate_logits), input_dtype)
         gating_scalars = xp.reshape(
             gating_scalars,
             (x.shape[0], gate_scalar_source.shape[1], self.lmax, self.channels),