So, if the base protocol may, given consensus, be upgraded at some point take on the additional decimal places to accommodate settling LN millisats to the blockchain, we'd have the option at that point of extending the halvings as well.
True. It also depends on when the additional decimal places are added. After the 10th halving, the block reward will reduce from 0.09765625 BTC to 0.04882812 BTC, with a rounding error of 0.5 satoshi being lost. This error is compounded in all future halvings, and added to with further rounding errors in subsequent halvings.
If we added an additional 3 decimal places before then, then we would end up going to 43 halvings instead of 33, and the total mined BTC would be 20,999,999.99997060000, for an increase in 0.0230706 BTC.
If we waited until just before the final currently planned halving, when the block reward was 1 satoshi, to add another 3 decimal places, then we would add 9 instead of 10 additional halvings, and the final mined bitcoin would be 20,999,999.97898740000, for a difference of 0.0020874 BTC more than planned, but 0.0209832 BTC less than the former scenario detailed above.
If we added 3 more decimal places somewhere between these two extremes, then obviously the final amount of added bitcoin would lie somewhere in between ~0.002 and ~0.02 BTC.