[FEATURE REQUEST] Support loading shared lora without clip input

[panxiaoguang]

Node Description and Use Case

We need a node called "LoraLoaderModelOnly" in ComfyUI, It's typically used in SD3 or SD3.5 when applying lora, because usually the Basic checkpoint don't have clip model, we don't need to input any clip in lora as well. But at this moment the shared lora loader have to input clip So how about adding a new node.

[doombeaker]

Sure, as SD3.5 doens't support lora now, we will support this node with supporting lora in sd3.5