Looks not bad to me. Could link or post an image with the symbol before? Would make it easier for me to compare.
Cemetery is not very visible but recognizable. How does it look currently?

I have updated both images with before/after renderings.

I'm not sure now about cemetery. With new symbol the lightness is similar and while I can spot the holes with general symbol with normal size (100-120%), cemetery symbol holes are not visible at 100% - it needs a lot of zooming (for me it's >200%), which is not what typical user would do. So maybe cemetery symbol should be moved to SVG too.

The general jewish religion symbol „after“ looks much clearer to me than the „before“ version. It can be recognized as Star of David.

About the cemetery symbol: Honestly, both, the „before“ and the „after“ variant, I cannot reliably recognize them as Star of David. They are just to small. But the „after“ variant – even though stronger – seems to me easier to guess as Star of David.

👍

Although i would assume this is clear but just in case it is not - this demonstrates the incorrect but consistent case described in #2750.

Sorry, I don't understand it - could you describe in short words what exactly is incorrect and what do you think is consistent with this case? Any ideas what should be done to make it OK for you?

The results in your After example above are not a faithful rendition of the SVG file, you can easily see this by comparing these results with a rendering of the pattern with other programs (Inkscape or ImageMagick) or with Mapnik when using polygon-gamma: 1;.

Incorrect but consistent refers to what i wrote in #2750:

It affects patterns in landcover-area-symbols - if observations in kosmtik are representative patterns in the normal landcover layer will render incorrectly but consistently with polygon-gamma other than 1 and correctly otherwise (the latter being the current situation for the quarry pattern) polygon-pattern-gamma does not appear to have an effect on this.

I don't know a way to fix this - otherwise i would have mentioned this in #2750.

It looks like people are in favour of this rendering not because it's rendered similar to a vector image, but because it works better (in a cartographic sense) than current PNG image for them - and for me too.

Solving the problem with Mapnik might give us the ultimate solution to this problem, but this PR would still make a progress with readability and fixes problem with map scaling, so it's worth merging IMO.