[Optimization] Support FA2/FA3/FA4 with attn_mask_q (#6354)

* support FA4 sm100

* flash attn backend support mask

* flash attn backend run flashmask correct

* add test for flash_attn_backend and flash_attn_func

* check

* add test for fa4

* requirements.txt add fa4 whl

* check test on sm100

* fix CI conflict

* add enable_torch_proxy for flash_mask

* lazy import fa4

* check

* fix tests import

* check test_load_mpt import

tests/layers/test_w4afp8_moe.py

@@ -6,6 +6,9 @@ import unittest
 import paddle
 from paddle.distributed import fleet
 
+# from fastdeploy.worker.worker_process import init_distributed_environment
+from utils import OpPerformanceTester
+
 from fastdeploy.config import (
     CacheConfig,
     FDConfig,
@@ -19,9 +22,6 @@ from fastdeploy.model_executor.layers.moe.moe import FusedMoE
 from fastdeploy.model_executor.layers.quantization.w4afp8 import W4AFP8Config
 from fastdeploy.scheduler import SchedulerConfig
 
-# from fastdeploy.worker.worker_process import init_distributed_environment
-from tests.utils import OpPerformanceTester
-
 paddle.set_default_dtype("bfloat16")