Is bounding_box setting respected with multimodal parsing?

[stonesthatwhisper]

I need to parse the tables in a PDF big in dimensions. Since the default settings returned nothing I am trying multimodal parsing.

1. Is the bounding_box argument to the parser respected in multimodal parsing?
2. Is parsing instructions passed to the vendor model too (so it might be worth the effort to adjust the parsing instruction.)