docs(docs/LLaVA_Video_1003.md) update: Fix Inference demo code

[dddst123]

Fix code in Inference demo: load_video method signature and parameter type

Problem: In the demo code for inference provided in the document, the load_video function does not require a self parameter; the max_frames_num parameter should seemingly be of type int, but the string type defined here will result in an error.

Fix:

• Corrected load_video method signature by removing incorrect 'self' parameter
• Fixed max_frames_num parameter type from string "64" to integer 64