Sentry mode requires the normal processing on the car to be active which I believe is about 300W. Not horrendous but over a 24 hour would give 7.2kWh, approximately an hour of charging from your charge at full power. This would be about 10% of charge per day lost, which seems a little high compared to what other people are saying about the phantom power loss.
That last 1% of charge takes a-g-e-s!
This is most likely due to BMS cell balancing, a well known 'taking ages' thing for packs - that does actually consume a lot of power. I don't know the innards of the M3 BMS, but here's how a few of the BMSs I have worked with operate:
Briefly as I can, a number of cells in series (they are also in parallel but let's skip that for simplicity) are charged with a voltage for the pack. The maximum voltage per cell is the voltage you really must have as a hard stop, otherwise cell damage will occur. Imagine 10 cells in series and we'll call the hard stop 4.1 volts. All cells get 41 volts charge voltage put across them and in an ideal world this would give us 4.1V per cell. In the real world some cells may have a little bit more voltage across as they don't always charge & discharge at the same rate. As they charge up we can end up with something like
Cell 1 to 9 = 4.0V and Cell 10 = 4.1V (in reality the difference is usually low millivolts)
with 41V applied across all cells this would start to put Cell 10 over 4.1V
The Pack has a board in it that senses the Cell 10 is too high so it drains power away, balancing the pack and allowing the others to charge safely. As the cells start to reach the threshold of 4.1 V more of them are dumping power to avoid overcharging.
The net effect is that it takes a long time to equalise the voltage across the cells and a lot of power is wasted in this phase of the charge as it is being dumped. A BMS that tries to recover this lost power is complicated and expensive, so in reality this balancing phase is seen as something that should be done on occasion and not all the time.
I've no idea how Tesla managed this in their BMS, though things like this are usually take care of automatically so we're not left having something else to worry about.
The last bit of charge gets progressively slower and power delivered by the charger vs power gained by the pack means it is not efficient compared to the lower SoC & faster bulk charge.
Charging will get throttled back if the cell temperature gets too high (or is too low) but Tesla have water-cooled packs so you may find pre-heat or cooling kicking in which, at low-ish charge kWh rates, may end up being a high proportion of the overall power used. IF a bucket has a small hole in the bottom filling it up with a tap isn't a huge issue, filling it up with a teaspoon at a time may require much more than one buckets worth of water. (Dynamic Equilibrium)
Final word is that the circuits in the packs that monitor the cell voltage can only handle dumping relatively small amounts of power, The balancing cables I have seen in most packs are quite thin, this really is a take it slow task and another factor in the charger power reducing as SoC increases.
In conclusion - I think you're alright, just keep an eye on it