Returns set of vectors, indexed by dimension containing interesting float
values.
The tests do not do the simple option for coverage of computing the cartesian
product of all of the interesting float values N times for vecN tests,
because that creates a huge number of tests for vec3 and vec4, leading to
time outs.
Instead they insert the interesting F64 values into each location of the
vector to get a spread of testing over the entire range. This reduces the
number of cases being run substantially, but maintains coverage.
Returns set of vectors, indexed by dimension containing interesting float values.
The tests do not do the simple option for coverage of computing the cartesian product of all of the interesting float values N times for vecN tests, because that creates a huge number of tests for vec3 and vec4, leading to time outs.
Instead they insert the interesting F64 values into each location of the vector to get a spread of testing over the entire range. This reduces the number of cases being run substantially, but maintains coverage.