delete flash attention

remove flash attention so the model can be integrated in mRNABench (has triton, which can't be used with flash-attn with this model)

bert_layers.py

@@ -24,10 +24,7 @@ from .bert_padding import (index_first_axis,
                                             index_put_first_axis, pad_input,
                                             unpad_input, unpad_input_only)
 
-try:
-    from .flash_attn_triton import flash_attn_qkvpacked_func
-except ImportError as e:
-    flash_attn_qkvpacked_func = None
+flash_attn_qkvpacked_func = None
 
 logger = logging.getLogger(__name__)