Proposal: omit layer surfaces from screen capture without blacking them out

[pikyusufaslan]

Niri currently offers block-out-from, which is right for sensitive content: the protected surface is replaced with a black rectangle in captures. There is a separate use case where that rectangle is not acceptable: a layer-shell recording/presentation guide should be visible on the physical output but absent from the captured frame, leaving the pixels behind it intact.

Examples: temporary recording HUDs, click/zoom guides, presenter aids, and accessibility helpers.

Would a layer-rule property for this make sense? For example, a surface marked for capture omission would not contribute normal or popup render elements to ScreenCapture/screencast targets, while rendering unchanged on the physical output. This differs from block-out-from because it omits the surface rather than substituting black.

I have a small local proof of concept that skips a dedicated layer-shell namespace during capture. I am opening this before a PR because the right rule name, target semantics, and interaction with popups should be agreed first.

Manual test target: show a transparent layer-shell guide above arbitrary content; verify it remains visible locally while a wlr-screencopy recording contains the unmodified content below it.