Requirements that are "not applicable"?

[m-mohr]

I've stumbled across a weird thing in the ESA Sentinel-2 L2A assessment requires the following:

The metadata identifies pixels for which the per-pixel tests (below) have not all been successfully completed.

The assessment from ESA is:

Not applicable to Sentinel-2 data. The test algorithm always provides a value and assign pixels to one of the following classes: dark feature, cloud shadow, vegetation, water, cloud low/medium/high probability, thin cirrus, snow, cast shadows, desert, no data, saturated or defective, unclassified

While it is fair that this doesn't apply, as a user that is using this product, how do I know this? It seems the only way to know this is reading the assessment? It's not specified on the DOI landing page and seems to be only clearly mentioned in the self-assessment.

Either the peer-review failed here or the requirement is not a proper requirement?! How can this be understood?

In the STAC extension I translated requirement 2.3 to the following:

There is a mismatch because the ESA data doesn't fullfil that, but I'm also not sure how I could better describe it. Should all requirements be phrased: "If applicable, then X is required"? That seems rather weird.

[mattsymbios]

Will forward this to ESA colleagues and GA peer review team for comment

[strobpr]

This seems a misunderstanding on the side of ESA filing the application. Of course the criterion is applicable to S2 data and it is fulfilled by including the class 'unclassified' to their value range of the per pixel mask.