I am a late comer to this thread and have not read all the posts. So, I apologize in advance if people already said what I am about to say.
1) Students learn about the Bohr-Sommerfeld quantization rule (BSQR) from the old quantum theory. They can not afford to “unlearn” the BSQR, as it is still extremely useful when the problem is too complicated. We are still using BSQR to quantize time-dependent solutions in field theory, to tackle the bound-state problem in QFT, to quantize the electric charge in field theory with non semi-simple gauge group, etc.
2) We perform the classical limit on the dynamics and operator algebra, but not on the Hilbert space \mathcal{H}. Itself, the Hilbert space is not an algebra, so we can not assign any meaning to the limit \lim_{\hbar \to 0} \mathcal{H}, even though the classical limit of the operators algebra in \mathcal{H} is a Poisson-Lie algebra.
3) In QM, the canonical commutation relations are nothing but statement about the homogeneity of the physical 3-Space.
4) There is nothing “wavy” in the Schrodinger equation i \partial_{t} | \psi \rangle = H | \psi \rangle. Waves, such as sound and EM waves, carry energy and momentum, while probability-wave (i.e. the wave function \langle x | \psi \rangle) does not carry energy or momentum.
5) Concepts such as “Particle Hilbert Space” and “Wave Hilbert Space” have no precise mathematical meanings. If you argue for “Particle Hilbert Space”, one can provide a “better” argument for “Wave Hilbert Space”: Without the Superposition Principle (i.e. waves and interference), “Hilbert Space” can not be linear vector space. However, even this argument is meaningless.
6) The axioms of any physical theory consist of two parts. There is (A) the abstract mathematical part, and (B) the physical part which maps the abstract entities introduced in (A) onto observations. Take QM as an example. In (A) we introduce \mathcal{H}, subsets of \mathcal{H} representing pure states, the algebra of bounded operators \mathcal{B}(\mathcal{H}), etc. While in (B) a) we talk about state-preparation procedure, b) introduce the average-value of the measurements of the observable (represented by self-adjoint operator) in the pure state (represented by the one-dimensional projection operator | \psi \rangle \langle \psi |), c) postulate the possible outcomes of a measurement and the frequency with which they are obtained, etc.
7) The classical notions of particles and waves are neither defined nor derived from the axioms of QM. The axioms are meant to describe (quantum) systems which we (the unfortunate creatures) are unable to perceive by our senses.