[MINOR] Supplementary-conditioner check uses identity (`!==`) instead of equality (`!=`)

[MINOR] Supplementary-conditioner check uses identity (!==) instead of equality (!=)

Summary

src/projected_to.jl:291:

if getconditioner(supplementary_ef) !== get_projected_to_conditioner(prj)
    error(...)
end

!== is identity comparison. It works for scalar (immutable, value-equal) conditioners, but for array/tuple-valued conditioners, two equal-but-distinct objects ([1.0] !== [1.0]) will spuriously trip the error even though the conditioners are equal.

Evidence (Julia 1.12.6)

[1.0] !== [1.0] -> true, whereas 2.0 !== 2.0 -> false. So a Laplace-style family with a vector conditioner supplied again as a supplementary distribution would be wrongly rejected.

Suggested fix

Use != (equality) instead of !==, or compare via a conditioner-specific equality that respects value semantics.

Start at src/projected_to.jl:291 and trace the supplementary-conditioner validation. Reproduce the reported case with equal but distinct array or tuple conditioners, then verify that valid value-equal conditioners no longer trigger the error while scalar behavior remains unchanged.