The TMOA reticle is only intended to be "accurate" at max magnification. At ANY other magnification will be off and math will be required to compensate.
It is also important to note Leupold
has had issues in the past figuring out which MOA they use. In my opinion, to be confident in a Leupold, a person needs to measure the reticle click values AND measure the reticle values to see what is actually installed in the scope and IF they match. Most people know Leupolds have a poor track record of accurate clicks but the math gets thrown right in the trash when your have an MOA reticle and Shooter MOA turrets (with error). When the OP knows these things for certain from testing himself then he can decide what to do (ignore the error, record the offset in the APP or get a new scope).
All too often assumptions are made that longer ranges are just a click away, and they can be but only if the correct data is used and not assumed.
Hopefully the OP can get it figured out.
As ForgivenALL, BC needs to be trued more than velocity. The Magneto Speed has an error of up to .5% or 15fps in this case. Start with this as a given and adjust the BC (correctly entering G1 or G7 value) to more closely match your trajectory. Beyond 1000yrds the muzzle velocity will also need to be slightly tweaked.
The OP has not said which bullet/BC combo he is using but I am sure if he does plenty will chime in with data that worked for them. Some manufactures like Berger, Hornady and Sierra provide fairly accurate BC's that only need a little tweaking to compensate for muzzle velocity (BC changes in flight), muzzle break, barrel (bore size, rifling, chamber). Other brands need major adjustments to compensate for the optimistic marketing of BC's
As the OP is now aware, the powder will produce different velocities at different temps. This temp data should be used to build separate "rifles" in his APP to be used in corresponding weather.
If all other data such as scope height over bore, actual twist rate (can vary) and bullets dimensions are input in the APP correctly then the last assumption to remove from the data is weather. I highly recommend the WeatherFlow wind meter for under $100. It works with several apps such as Hornady 4dof but I prefer the user interface of BallisticsARC (free).