Untested and dead code in daubechies_scaling.hpp

[jzmaddock]

Probably one for @NAThompson.

In daubechies_scaling.hpp

We have one line uncovered here:

math/include/boost/math/special_functions/daubechies_scaling.hpp

Line 367 in 2af21de

grid_refinements = r[p];

which appears to be because the constructor is never called in our tests with anything except the default value for grid_refinements. We should have at least one absolute error (-2) value, plus one other positive value. But I'm not sure which tests and invariants hold in these cases?

Also the types detail::matched_holder and detail::linear_interpolation appear not to be used by anything, nor mentioned in the docs?

[jzmaddock]

Sorry, one other, there are no tests for double_prime() other than the boundary cases. Again I'm not sure how to validate the result?

[jzmaddock]

And one more, daubechies_wavelet.hpp is also untested for anything except default argument to daubechies_wavelet. The code for p==1 here:

math/include/boost/math/special_functions/daubechies_wavelet.hpp

Line 212 in 2af21de

if constexpr (p == 1)

is also untested.

[NAThompson]

@jzmaddock : Yup, you're right. Will try to get to this next week.