Theta "drift"?

I'm noticing two issues that may or may not be related, but in both cases involve what I'm calling "theta drift".

Issue #1: When I first got my table, I played a track I had created in Sandify that contained text in order to determine the "top" of the Sisyphus canvas so that I could orient my table accordingly in my living room.  But I just played that same track again a few days later, and that track is now playing rotated 90 degrees clockwise from the way it played originally.  I'm not sure if this matters, but between the two runs of this track, I installed firmware update 1.10.70.  If that doesn't account for it, does it make a difference whether a track is played as a single track or played in a playlist immediately after a previous track, i.e. does the table only "zero/reset" the theta value it has accumulated from playing the preceding track in certain conditions?  Is there some way to make sure that non-algorithmic tracks will always play in the same orientation on the table?

Issue #2: I'm finding that the LED Calibration offset needs to be adjusted in order to stay precise.  Thus far I've had to use values ranging from -4 to -15 in order to keep the center/brightest portion of the LED tracking the ball as precisely as possible.  It's not a big deal since it's fairly close even when it's not perfect, but I was curious if this was expected.