The source product you have provided is almost complete out of your defined bounds and has only a few valid pixels. But I think this is only an example.
But when using your settings, I see the small overlapping part in the mosaic.
What kind of an issue do you refer to.
One issue I see is at the edge of the source image. Even better visible when adapting the mosaic area to the input scene.
Up to a certain distant the closest pixel is searched, and this leads to a duplication of pixels. The searching is necessary for the pixel-based geocoding.
This has been observed before.
- OLCI L1B reproject artifacts (with PixelGeocoding) - s3tbx - STEP Forum (esa.int)
- Reprojection output extends beyond edge of data - snap - STEP Forum (esa.int)
The second one contains a workaround. But this would mean that you need to do the l2gen processing again.
You cut out a larger area, do the l2gen processing on them, and then reprojecting the scenes and cut out the final region.