Review uses of Jasmine toBeCloseTo

[javagl]

Some specs use the Jasmine toBeCloseTo function to check whether an actual value is close to an expected value. The function receives a precision parameter that is documented as "The number of decimal points to check". This is used internally, roughly: as an exponent of 10, to basically "shift" the values into the right range.

Some of the specs seem to assume that this precision parameter is supposed to be an "epsilon". One example (just a random search result) is here, where CesiumMath.EPSILON10 is passed in as the 'precision'. This essentially acts like an epsilon of 0.5 (!). Specifically, a test like

const expected = 0.1;
const actual = 0.3;
const EPSILON10 = 0.0000000001;
expect(actual).toBeCloseTo(expected, EPSILON10);

will pass, which is certainly not expected.

Other places seem to use the precision parameter properly (e.g. with values like 2 or 3).

The Testing Guide recommends to use the (Cesium custom) matcher toEqualEpsilon. I think that in many cases where someone passed in some epsilon to toBeCloseTo, this call can just be replaced with toEqualEpsilon. But where exactly the wrong assumption about that parameter was made has to be reviewed.