Avoid special casing the dot product in CliffordNumbers.mul

[brainandforce]

In order to correctly calculate the dot product of two multivectors, there is an if statement to check if the CliffordNumbers.GradeFilter argument is the one used for the dot product. This is used because the dot product requires both masking the multiplications of certain grades and altering the signs associated with each multiplication:

# Special case for dot products
signs = sign.(inds)
if F <: GradeFilter{:dot}
    signs = signs .* Int8(-1).^(grade.(inds) .* (grade(a) .- grade.(inds)))
end

I think we should avoid specifically referencing any CliffordNumbers.GradeFilter instance in CliffordNumbers.mul. Perhaps we'll need an additional function to handle the sign manipulation.

[brainandforce]

Fixed with 7cb5747