Maybe I'm thinking of this too much in terms of vectors like pool balls headed for the rail, but surely there must be some extreme angle where the direct sound just glances off the boundary and doesn't get the +6 dB boundary effect, right? And if that's true (IF!), does the angle gradually increase the boost up to +6 as the angle moves from that extreme to being perpendicular to the direct sound source?
Not sure the pool ball analogy is apt here, but pool balls still behave as vectors at even quite shallow angles to the rail.
I'll relate a situation from my own experience where the sound sources were predominantly positioned at very shallow angles to the boundary- In combination with a bunch of other mic techniques, I used to boundary-mount miniature omnis on a hard wooden stage surface for a series of non-PA amplified jazz trio gigs. In that case the mic(s) were placed on the wooden floor (a hard boundary facing upwards) in front of the band, and the guitar amp, drum kit and small acoustic bass amp were sitting on the floor at various distances from the microphones. As seen from the mic(s), the uppermost edges of the more distantly placed guitar and bass amps had to be no more than 5 degrees above parallel. The drums in the kit ranged from 1 or 2 degrees (bottom edge of the kick drum low to the floor) up to 30 degrees or so above parallel with the cymbals up to maybe 45 degrees or so above parallel. The acoustic bass was closer than its amp and the bass body probably ranged from 10 to 45 degrees or so above parallel. Except for the occasional guest sax player and stage banter from nearly directly above the mics, all sound sources were at the very least 45 degrees away from perpendicular, with the majority of them much farther away from perpendicular. The sound was very clean and clear for all sources. I was especially astounding by how clearly the guitar amp came through, which was the furthest source 15 feet away or more, and I remember thinking that it sounded like the sound almost flowed more easily across the floor than through the open air. That guitar amp was basically at a near right angle to the boundary surface.
I really liked placing the mics (not just boundary mounted, but more typical stereo setups) very close to the floor partly for this reason.
My seat of the pants take away from this, from other boundary mounting experiences, and from experimenting with close mounting mics to the surface of Jecklin disks and baffles that have soft fuzzy sound absorbing surfaces, is that the surface of the boundary matters and is likely to be the largest contributing factor for far off axis angles.
If the surface is hard, smooth and reflective (which it is supposed to be for proper boundary mounting effect), source angle away from the boundary plane doesn't seem to matter much if at all. If the surface is soft, fuzzy, furry, and sound-absorbing, the behavior is different. The highs are reduced, and the response seems to change more as the source angle gets close to parallel. Not overly surprising I suppose.