I will venture to guess that this has to do with stopping on a switch. First, it is not a great practice to be stopping on a switch in order to pick up or drop off passengers, but second and more importantly, detecting the presence of a stopped train on a switch is to say the least cumbersome and more error prone for the signal system. I am not sure if detection of a stopped train on a switch is done even in places such as Penn Station or Hoboken Terminal. Given that Harriman is one of the stops with the largest number of riders, all trains for the foreseeable future will stop there -- the extra hundred or so yards of track did not give any advantage over the shortened configuration, so what was the point of having them? If I have to guess, the switch was there first and the station came second. The stopping on the switch was tolerable before PTC, but now the trouble is just not worth it. The platform configuration was probably not a consideration at Harriman.
Otisville appears different. The trains do not stop on the switch and given the low ridership (lowest ridership on the line), the platform configuration was probably not deemed an issue.