I have no idea if their engine can support this or not, but I think the flow of cinematics is important in their game but having a fuller, more correct interpretation of DnD5e is wanted by the player base.
One way to help resolve the differences between the two desired outcomes is to implement the reactions in such away that no animations are performed until all actions/reactions have been selected but the ruleset happens, doing all the actions/reactions of an actors current action. Once all completed, the game will play all the required animations needed to play out what happened in the rules. Example:
Paladin A, controlled by Player Z, wants to move and attack Goblin G. (Player Z clicks next to Goblin G) - Game plays animation of Paladin A moving to Goblin G. - Player Z chooses to attack Goblin G. (Player Z clicks attack on Goblin G) [Game internally does a roll for the attack roll vs ac. It displays a 19 + 3, scoring a critical hit. This is shown in the combat window.] Player Z receives a reaction window saying "Do you wish to use Divine Smite?". (Player Z clicks "Yes") [Game internally does the damage roll and shows the current attack damage roll.] - Game plays animation of the Paladin swinging their sword, flashing into a glow and smiting the goblin -
Basically, the game plays in the background in terms of mechanics and if reactions are needed, the game prompts the required actors (AI or player) for the needed inputs. Then, then all the needed inputs are done for the action being performed (and the reactions that will happen from the chain of events, like an action that allows a reaction that might set off another reaction), the game will play the animation based on the inputs. This allows the game to play the animations smoothly all in one go based on said action/reactions.
The other way to achieve this is to allow reactions to stay automatic like now, but allow the player to setup a FSM (finite state machine) for the various actions that you want automatic. If the reaction doesn't meet the requirement set, then the reaction won't trigger. This does mean reactions will be tedious to setup in the beginning and might need to be changed from time to time dependent on the fight, but would achieve the same outcome of not interrupting animations.