Perimeters are generated hanging/floating and not detected as such

[PreownedFIN]

Description of the bug

Quite similar to #12670, but slightly different, as these extrusions do not connect to any other extrusions (except above them).

Perimeters are generated hanging, and are not detected by bridging perimeters, nor by "hanging perimeter"-checks that PrusaSlicer does.

Perimeters generate this way with high layer heights with Arachne, and not exactly the same with the classic perimeter generation (though classic has its own bigger problems).
The model inside the project file generates these on all nozzle sizes when the layer height is 0,3 mm or greater on the Prusa-supplied MK3 profiles of all nozzle sizes, with the Prusa supplied >=0,3 layer height profiles (with detect bridging perimeters on or off).
I noticed this on the 0,8mm nozzle MK3 profile and 0,4mm QUALITY profile, but checked on other nozzle sizes and layer heights, and all generate suspect perimeters.

To be fair, these are not bridging perimeters per-se, just perimeters that have no connection to other extrusions in the same- or previous layers, just future ones.

This is an especially difficult part of the geometry for the perimeter generation to get consistent, with wide extrusion widths on top of that.
But IMO at least the print stability checks should notice these floating extrusions in some way or another. (Within reasonable computational time ofc.)

Project file & How to reproduce

Open the attached project, and slice the model with any print setting with =>0,3 mm layer height.

bridgingProblem.zip

Checklist of files included above

• Project file
• Screenshot

Version of PrusaSlicer

2.8.1

Operating system

Windows 11

Printer model

Prusa MK3@0,8 mm nozzle

[u89djt]

(Fellow user following the hub)
Re detecting problem extrusions, consider this situation:

It's your job to provide support when there isn't a surface available to recognize as shallower than some limit.

So you either support All The Things

Or do some painting

Painting won't help with your model there because most of the volume of the sharp section below the extrusions is physically impossible to print, and it'll try to support what you painted, which will miss the fragments of extrusion.
The slicer can't tell what's happening with those tiny triangulated regions. See how it has to guess and miss:

When you let it have a go at supporting it, it follows the thing it's supposed to be rendering: the model.

You might ask why supports aren't generated to meet the extrusions rather than the model.
To do that, you'd have to understand the relationships between the extrusions to keep track of what is supporting what already. Those relationships are embodied in the original model. The extrusions are just squirts of plastic. So you definitely need to refer to the model, and you hope that the plastic is a sensible rendering of the model.
I guess you'd have to rebuild the extrusions as a mesh model and analyse that for supports when the model isn't actually printable in places.

For your model, you'd have to ask the slicer to beef up the sections that are too thin, and it would have to guess in what directions (you can see it failing to guess already), so you need to change the model to be able to print it. The slicer doesn't see what you see.