A counter-argument to this can be made using the D major prelude from WTC Book II (BWV 874), see http://imslp.org/wiki/Special:ImagefromIndex/02182
Here there is a double time signature, "cut time" (Alla breve) followed by 12/8, and the opening bars are (apparently) alternately in 12/8 and cut time - though there is an unanswered question as to why the time signature puts the cut time first, but the music starts with the 12/8 rhythm.
After the double-bar, the "12/8" and "Alla breve" motifs are combined in counterpoint. But as Tovey said in the notes on his edition, if the combined passages were meant to be played as in the OP's reference, it seems incomprehensible that either (a) if the whole piece is meant to be in 12/8, why did Bach go to so much trouble in the first few bars to write one simple rhythm incorrectly in terms of a different one? or (b) If the opening is in cut time but the later passages are not, why does the "cut time" motif inexplicably change its rhythm when it is combined with something else?
(Note: that is my paraphrase of Tovey's comment, not an exact quote - I don't have his score available right now).
With "two beats in the bar", the prelude makes good musical sense played exactly as written.
A possible explanation might be that this elaborate notation was really an exercise for a pupil about reading French-style "notes inégales". Personally I'm not very convinced by that - Bach's "notes inégales" movements tend to be written in dotted-8th plus 16th-notes (e.g. Contrapunctus II of the Art of Fugue).
I don't think there is any debate that if a triplet is written against a dotted-8th plus 16th, as in the OP's second example, the meaning is that the 16th-note is either 1/3 or 1/6 of a quarter note, depending on the tempo (i.e. it is not a mathematically exact 16th-note). Dotted rhythms were always open to interpretation as to the exact length of the dotted notes.
EDIT: The OP's first example (BWV 608) shows a modernized version of the rhythm as originally written by Bach. You can see Bach's autograph here: http://imslp.org/wiki/Special:ImagefromIndex/111276
Bach wrote the triplets "incorrectly" as 8th-notes, not quarter-notes. On the middle system of the second page, he writes a few "quarter + eighth" triplets explicitly as such. If you want to interpret all the duplets as triplets, this raises the same question as with BWV 874: why did Bach write two different notations for the same rhythm?
The form and instrumentation of the piece seems relevant here. It is a chorale prelude in 4 parts, for organ. The tenor part is marked "Ped", meaning it is to be played on the organ pedals. The autograph writes the tenor at the sounding pitch of the notes. But that is only half the story, because the piece is acutally a double canon. The chorale melody appears as a canon in the octave between treble and tenor, and the alto and bass are another canon at the octave. Both canons continue till the final few bars.
It follows from that (and the argument is supported by the frequent collisions between the long notes in the treble and short notes in the alto) that both the treble and tenor were meant to be played on the organ pedals using a stop that sounds an octave higher than the standard (8 foot) pitch. Thus, there are four independent parts, each played separately by one hand or one foot. In the final few bars, the treble and alto parts cross over, otherwise the top part goes too high for the organ pedals. The organist's feet are then playing the two inner parts of the four part counterpoint.
Of course this is the sort of "Cameron Carpenter" showing off that got Bach the reputation for making church music sound like a comic opera, to the bewilderment of the congregation and the disapproval of the church authorities, but that's a different story... (If you never heard of Cameron Carpenter try this for starters:
If this (lively and exuberant) piece were performed by four soloists (e.g. two trumpets and two trombones) the most natural way to approach it would be with an even (and fairly quick-paced) 3-in-a-bar beat, and to let the subdivisions of the shorter notes fall where they may. If the individual players want to improvise a bit, well, let them do that!
An organist (with enough technical skill) can do the same playing all four parts at the same time, just as a jazz drummer can play four independent rhythms at once while maintaining a regular "beat". But if you start to study the score from the point of view of a pianist (and there is no particular problem playing "all the notes exactly as written" on a piano) it may not be obvious what was originally intended.