StellaVLA: In-Context Structured Demonstration for Generalizable Vision-Language-Action Models | Startup Networx | Startup Networx