The tilt is to be able to send a bit more high frequency energy into the room without having it sent directly at the listener, as well as getting a more even vertical dispersion / similar frequency response for people sitting as opposed to standing in the room. Remember that these speakers are designed to be parallell with the wall (not angled towards the listener), so we need some extra energy in the high-end to get a good response off-axis. In summary this gives the most even response at the listening position as well the best and most favorable compromise with regards to response / dispersion patterns wherever you may be in the room. It's basically tonally similar anywhere in front of the speaker, until you actually pass by on the side of the speaker (so more than 90 degrees if possible in your room), where you get an interesting effect of rapid midbass roll-off due to the cardioid pattern.
With regards to dips: As mentioned these are in-room responses (listening position), so all dips are room related. The anechoic response is +/-2dB across the range, and that the deviation is even that much is mostly due to by design lift in the low end, and by design recessed response in the 2-4khz area.
Here's nearfield 15deg off-axis 400-20,000hz (below that you'd see the room affect the response as this isn't measured in an anechoic chamber) - it's basically +/-1dB except 2-4khz (on purpose), and would be the same in the bass in an anechoic chamber.
View attachment 58070