I got those made here: http://astraproducts.com/ asking for hardcoated (scratch-prevention) filters with a 0.5mm thickness and minimum 90% transmittance between 930 to 950nm. They haven't actually arrived yet so I will let you know if it actually works. But going by the datasheet of the VL53L0X, it should meet the standards. Also, I have seen another project where people also used acrylic glass for the VL53L0X, so it definitely should be possible.
The sensors still work fine. After I took them off and inspected them, they had tiny water droplets (or just moisture) on them, blocking the signal. One wipe was enough to make it work again. I imagine though that once you pour enough water on it, they should eventually stop working, so I want to be careful with them now. Also, as winter is approaching I imagine the freezing and thawing of water would have a destructive effect on them too.
The 10 sensors are used by the trash company yes, but more as in a pilot study. The accelerometer looks very good, and when the lasers work they do give fairly accurate data too. Once I can implement some improvements on them we will probably get more and use them for real.
The JSN-SR04T I haven't actually used yet in the field, as the measurement angle is so ridiculously large (something around 50-55 degrees when I self-measured it) I favoured the VL53L0X.