fix: correct text input validation logic in 8 multimodal processors (and → or) (#47663)
fix: correct text input validation logic in 8 multimodal processors
Replace logical 'and' with 'or' in the isinstance guard:
elif not isinstance(text, list) and not isinstance(text[0], str):
The original condition only raises an error if text is BOTH not a list
AND its first element is not a string — which is impossible since a
non-list cannot be subscripted. The correct check should raise the
error if text is not a list OR if the first element is not a string.
Affected processors: blip_2, gemma3, gemma3n, idefics2, instructblip,
instructblipvideo, phi4_multimodal, qwen2_audio