Improve Architecture Documentation

[placerda]

Provide diagrams that show which modules to it are optional vs. required.

For some of our accelerators (AI LZA/GPT) they have diagrams that are quite large. These diagrams depict the "if you installed all modules" and doesn't really show what pieces are optional.

Often times it causes confusion and makes the field/customers think it is overkill, or costly for no reason, and makes conversations difficult trying to highlight what their choices are. Moreso if it's still being considered to nominate, it stops the field from even submitting.

[JiwaniZakir]

The core issue is that the architecture diagrams currently represent a "maximum deployment" view rather than a modular/layered one, making the system appear more complex and costly than it may need to be for a given use case. This is a common documentation anti-pattern where a single monolithic diagram is used instead of tiered views (e.g., a minimal baseline diagram plus optional add-on overlays). A likely root cause is that the diagrams were originally authored to showcase full capability rather than to guide decision-making during evaluation or sales conversations. Introducing a matrix or legend distinguishing required vs. optional components — alongside a simplified "minimal viable deployment" diagram — would directly address the confusion without requiring changes to the underlying architecture docs structure.