Do you think that art can actually be a mode of access to divine transcendence?

Yes, but without constraint, without injunction. 

By internal pathways?  This is the case with Messaien?

One is not obliged to share the motivation behind the composing, even if there is for the creator a complete adherence of his motivation to his composition.  And the music lover (amateur) is not required to repeat his path.  Adherence here .is not in the order of  imitation of his motivation.  Let us take an in-between case: the Masonic tonality of Figaro. One is in no way forced to share in this rationalistic religiosity and  follow a path that is not his own.  I take this example from a theologian I admire, Karl Barth, who would put Mozart above Bach.  Bach was intentionally religious, Mozart not.  But one can listen to Mozart with a fervor which will reveal religious motivations in us.  In a certain way Bach constitutes the threshold of religious aesthetics, bridgeable or not.  Kant himself had admitted another threshold, that of  ethics by way of the sublime.  In the sublime, our imagination is overcome by excess, quantitative or dynamic, but we are sheltered, that is we reaffirm our moral superiority faced with the superiority of those forces which would destroy us if we were delivered over to them.  But one can also say that an ethico-moral tonality is evoked by the “Starry Sky Above.” The sublime also has a potentially religious power, but not expressly, not necessarily.

Vladimir Jankelevitch notes, in regard to music, another form of threshold:  that of the mystery, of the inexpressible, which relates to the endless, untiring, inexhaustible work of language to say what cannot be said or said only by allusion, suggestion, allegory, or metaphor.  The musical mystery, he writes, is not the unutterable but the ineffable.  It is the black night of death that is the unutterable, because it is impenetrable shadow and despairing non-being, and because an unbridgeable wall keeps us from its mystery:  it is unsayable in this respect, that there is absolutely nothing to say, which renders man mute by crushing his reason and  paralyzing his discourse.  And the ineffable, quite the reverse, is inexpressible because there is something to say about it  interminably, infinitely: such as the impenetrable mystery of God, or  the inexhaustible mystery of love, which is the poetic mystery par excellence.  Do you think that art is a way of reaching this frontier of the unsayable or the ineffable, death, love, the mystical experience and perhaps still other similar regions, which would underscore the uchronic and utopic function of art?  

The ineffable has a character of incohesiveness, of indifferenciation which is precisely surmounted by the work of art.  This latter is structured otherwise than in language, but it is structured; and in this sense each work of art has the singularity its structure.  In the pages I devoted to the aesthetic experience at the end of Critique and Conviction, I insisted above all on this singular structured character, the fact that each work is the resolution of a problem..  Here one could take up again the analyses of Merleau-Ponty on Cezanne. In painting the problem itself is singular: it is the conjunction, in the same undertaking, of  color, form and light, and this combination is singular each time.  What would appear to me to be ineffable, I would put not in each painting, but in what has provoked it, namely, if  we take the example of Cezanne, in this permanent return to the object of painting, as if there were something inexhaustible in the saying.  There is a kind of tenacious approximation, in favor of another perspective, another profile, different each time.  Thus the signified “Mount Saint-Victoire, if we might say so, is a demand to signify more.  I would insist here on the ineffable command and the effectuation that is singular each time.  This is an analysis that I found admirably done by Granger with respect to the algebra of Pascal.  The proper name is the name of the singularity of the resolution of the problem.  Here we find the initial affirmation:  this singularity of the resolution of a problem, which brings a singular response to a singular challenge, is eminently communicable. We compensate for the lack of universality of the singular resolution of the problem by the communicability.  Here there is evidently a parallel with Kant when he underscores that it is the game of understanding and imagination that is communicable. And in the case of the solving of a problem, we can say that it is the game of challenge and solution.

We can also understand something else in what you have just said with respect to Cezanne. In fact, what is this need to take up again ceaselessly these approximations of the painted object?  Is it not the question thematized by Husserl of the flux of  Abschattungen,  of faces, sketches, profiles, silhouettes in a temporal horizon of perception? The work of art, then, would it be in Husserlian terms rather on the side of the noematic correlate, on the side of the transcendental, object, or on that of the noesis, on the side of the intentionality of the subject; would it not finally be this relation between object aimed at and the aim of the object which could define the work of art?

I would prefer to broach this question by way of its linguistic equivalent, namely that a linguistics of a Saussurian type, which is binary, does not work.  Signifier and signified are the heads and tails of  the sign.  What is needed is a semiotic of three terms:  signifier, signified, referent.  This is the demand for a referent that is not met by the binary signifier-signified.