TGridPrecisionLimit< FRealSingle >

Precision limits where our floating point precision breaks down to 1 (each value above this adds more than 1 to the number) Note the F32 limit is actually higher than this point (should be 8.388e6) but kept at 1e7 for legacy reasons