v1.0.34

Returning to float32 precision from float64.

When I previously tried this, my export results were perceptibly slightly off compared to SAI. It turns out that this was caused by using the float32 epsilon value when dividing in blend functions, which instead now just checks if the divisor is zero. Fixing this allows exporting to use float32, which is faster and uses less memory, while allowing the export to be identical to SAI's exports. There are still very tiny floating point errors that accumulate, but the differences from SAI are imperceptible.