What do you mean by "the problem is back"? It came back because of what?
The spindle control is not syncronous motion, it never was syncronous, because it is controlled by the M3 and M4 and M5 text macros which macros when executed the motion controller must flush all of the motion to finish them and then execute the macro on the PC side and then can continue the g-code execution. This is not a problem for a spindle control, but can be for a laser. But M3, M4 and M5 are not for laser control.
And this can't come back as a problem since it was always like this.
For laser there are the M10/M11 commands which are motion syncronous commands and which executes between the movements without pause.
If there is a decceleration at the end of the motion depends on the trajectory settings, the CV parameters.
For example if the connection angle is set to 89° in the CV parameters, but the connection angle is 90°between the 2 lines then the controller will deccelerate to the connection point.
If the angle is set to 91°with the same example then it will not deccelerate, but will blend the path using the CV parameters which can still mean some sort of decceleration if the feedrate can't be physically kept using the set parameters, but usually is will be then not a full decceleration, how much it will deccelerate and how the path can be followed depends on the actual g-code and the other CV parameters.