docs: document the start/end row in row proof

[rach-id]

Description

Adds a simple documentation to the start row and end row fields in row proof to show that they're not used for verification. But if anyone is using them downstream, they should validate them themselves

[rootulp]

I'm pretty confused by the comment. Any chance it can be rephrased to clarify what it means?

[rootulp]

[question] can we please clarify "it" here? Is this correct? "it" is used four times but I think they mean different things.

Suggested change

	// Note: it's currently not used for verification, only for validation.
	// If it's used downstream, make sure to validate it.
	StartRow uint32 `json:"start_row"`
	// Note: StartRow is currently not used for verification, only for validation.
	// Before using a RowProof, users are expected to call Validate().
	StartRow uint32 `json:"start_row"`

If not, can we clarify this comment in another way?

[rach-id]

4f4c617 ?

[rootulp]

Seems better to me.

1. What does "extra validation depending on how it's used." entail?
2. Why would this field be unused downstream? It seems like a necessary component of RowProof.

[rach-id]

Extra validation depends on how those fields are used.
If, for example, they're using that row range to specify where the data is. They will need to check if the start and end index are exactly the ones where the data is.
So, it depends on what downstream repos are proving and how they're constructing their proofs.
Ultimately, we might remove the two fields. But that's a breaking change, and it's not a priority.

When verifying the inclusion of a set of rows to the data root, you don't necessarily need to know where they are. What you care about is that they're committed by the data root. This is done using the slice of proofs and row roots.

[rootulp]

If, for example, they're using that row range to specify where the data is.

Does any downstream repo use StartRow or EndRow to indicate anything besides the row range where the data is? It seems like these fields should only be used for that. If they're present they describe the row range.

When verifying the inclusion of a set of rows to the data root, you don't necessarily need to know where they are. What you care about is that they're committed by the data root. This is done using the slice of proofs and row roots.

Got it, thanks.

[rach-id]

Currently, no. But when building zk-vms that create/verify these proofs, some teams asked about the fields whether they need any specific constraints. That's why we're clarifying that in the comment + letting them know that if they have something specific in mind, they should think about their constraints

[rach-id]

CI fixed here: #1474

[staheri14]

Added some comments/questions about some validation that can be relevant to the StartRow and EndRow.