It was a rough night in filament land.  Because the container contacts the window & the window cools to 40F at night, the container got much colder than the room.  The filament dutifully absorbed all the water in the container as shown by the dew point falling below ambient.  During the day, the container heated up.  The filament released its water as shown by the dew point rising above ambient.  The pump ran for its full duty cycle, then got shut off before it reached the goal dew point.  It ran all day but never at night.  Running the pump corrupted the serial port data, maybe because of voltage ripple.

Inside humidity never got above outside humidity, so dewpoint really was the only useful parameter.  

It wouldn't be a very useful container if the inside temperature didn't get above the outside temperature, but clearly touching the window dropped the inside temperature at night.

During the day, the pump finally blew off the lid & the dew points equalized.  So the mane problem is keeping the filament from reabsorbing water when it cools down.  Then comes keeping the pump from blowing off the lid.  

The container seems to be well isolated despite having no exit valve.  The pump is noisy.  There's a good argument for having a higher capacity blower force air in through a zig zag path.  A blower wouldn't develop enough compression to blow the lid off.  It would keep the dewpoint closer to ambient.  The closer it is to ambient, the more water it can pull out of the filament.  A higher resistor might also keep the lid on.

The only idea coming to mind for the reabsorption of water is using PG&E to heat the container when the dew point inside is lower than outside.  It's ridiculous to heat up a container to counteract being next to the window & it's not going to go well in the summer when the low temperature is 80F.  Filament is going to reabsorb water the moment it's taken out of the container & used for printing.