To start with, we need to understand what actually happens between ASIAIR and your mount during a Meridian flip.
The ASIAIR does not actually do a Meridian Flip slew (except for EQMOD, where the INDI EQMOD driver inside the ASIAIR acts as a mount).
What really happens is that the ASIAIR depends on the Mount to do the Meridian Flip slew.
In general, when you execute a GOTO to a mount, the mount uses the current Local Sidereal Time to determine if the GOTO Target is east or west of the Prime Meridian, and where it thinks the OTA is relative to the telescope pier (or tripod if that is what you use; it is called pier after real telescopes).
"German" mounts will always try to make sure that the OTA is on the West of the pier when the OTA is pointed east of the Meridian. Once the target moves past to west of the Meridian, if the mount is again commanded to GOTO the same object, the mount will reposition the Polar and Declination axes so that the OTA will be east of the pier. This is called a Meridian Flip. You can check this by going outside to look at your telescope: the RA axis will always be tilted so that the OTA is on the opposite side of the pier from the target.
On top of this, most mounts have what is called the Meridian Limit -- when the RA axis travels past this Meridian Limit, the mount will turn off tracking (to avoid objects hitting the pier or tripod, cable snagging, etc). The Meridian limit is often set to 0 degrees past the Meridian but with some mounts (like my RainbowAstro), you can change this limit number.
So, given this understanding of these two mount properties (i.e., (1) GOTO will position OTA on opposite side of the pier as the target, and (2) the Mount will refuse to track past the Meridian Limit), we can try to guess what is happening to you.
What happens during the Meridian flip is this (note that ASIAIR merely looks at the time and execute mount commands, it has no idea what the path the RA and Declination motors really need to do):
1) ASIAIR waits until it reaches the time A before the target reaches Meridian (on v1.5, A is settable by the user in Telescope Settings > Meridian Flip Settings > Stop Tracking x min before Meridian (by "min," ZWO means minutes of time -- not documented of course, what else do you expect?)
2) ASIAIR stops imaging, and takes a plate and solve it to find where the OTA is pointed, call this coordinate P.
Note that while it is taking the plate, ASIAIR assumes the mount is still tracking.
3) At some time B, the mount stops tracking because the RA has reached the Meridian limit.
4) Time passes until time C is reached. Time C is the second parameter that you can set in v1.5 ASIAIR as "Do AMF x min after Meridian."
At time C, ASIAIR assumes that the mount also agrees that the object has moved past Meridian. So ASIAIR issues a GOTO to point P (plate solved above).
Since the mount thinks that the object is now on the "wrong" side of the pier at time C, the mount will execute a Meridian flip.
ASIAIR has nothing to do with the actual flip motion. There is no magic in ASIAIR except to carefully choreograph the timing (time A and C).
You can also guess by now that timing is very critical. If your mount thinks it is 10 minutes (time) ahead of the ASIAIR, the mount would have stopped tracking a whole 10 minutes before ASIAIR thinks the target has reached Meridian.
Conversely, if the mount thinks it is 10 minutes later than ASIAIR time,then the GOTO in step 4) above will do nothing since the mount thinks the target has not moved past the Meridian yet.
If your mounts time cannot be sync'ed up to the iPad (I know people on FaceBook who refuse to sync; and if they cannot Meridian Flip, now you know why), I would recommend a large time for the two time parameters A and C. Use something like 3 minutes, if you think the Mount's time and the ASIAIR time can only be guaranteed to be within 1 minute of one another. If the times are in sync, you can use 1 minute for each parameter. Do not, under any circumstance use 0, since any quantization of the clocks will throw of meridian flipping, unless you change the Meridian Limit of the mount to set it a little past the Meridian.
Larger parameters will cause your system to wait before it resumes taking images. But it is better to lose a few minutes than to lose half your night of imaging.
You can also tune the Meridian Limit (usually expressed as angles, so you have to convert that to time) to ensure that time B is no earlier than time A. It can actually be later than time C, since the mount will still do a Meridian Flip while it is tracking. Normally you will choose a number for the Mount Meridian Limit to keep the OTA from hitting the pier.