Issues

• The PCB was a little too thick. Mine were 1.6mm nominally so better get 1.2mm.
• BME688 reads a temperature a few degrees above ambient. I was hoping the fan of the Vindriktning would cause enough air circulation to avoid this. I don't know if the sensor in the Vindriktning requires some heating like the SGP30 does and causes this or if it's just from all the electronics in such a small space.
• Sourcing SOT23-5 voltage regulators...
• Sourcing PicoBlade connectors. In the end the compatibles ones I ordered arrived also in time.
• CCS811 vs. SGP30: I was curious about the performance of both devices so I found this comparison. I think I also read somewhere else that the CCS811 should be more stable. However, my experience was the opposite, with the CCS811 constantly drifting. 
• The SGP30 seems to be very susceptible to humidity. After running the AC and opening a door or window the humidity increases and the TVOC/eCO2 values go through the roof. After a few minutes they seem to stabilize again.

Future improvements... maybe?

• Move BME688 to the outside or up against the perforated back wall of the sensor. Or just set a tempoffset in Tasmota.
• I wanted to modify the 3D printed mounts for the LED breakouts to be a mask showing the type of measurement, such that the LED light would project "CO2" or "VOC" in different colors through the Vindriktning front plate. Unfortunately, the writing would be too small for my printer, so maybe I'll have to order them somewhere or get a smaller nozzle.

This nicely shows the effect I was going for with the modified LED mounts.