The response and heat transfer is better if the heater/sensor/tip were all in one, so you don't have issues regarding parts fitting and integration. Also, the closer the heat sensor is to the tip, the better the feedback and thus more accurate the tip temp is, otherwise you have an 'offset' which represents a difference between the tip mass versus the mass of the base. You can compare responses above.
The question becomes is it worth the extra cost of the newer tips and the station? Clearly, every step in integration is better, between these and the oldest. But, then why not the JBC heating model? or the Metcal model?
I think as you move up, you have to justify the cost over productivity, so the heavier your soldering needs, the more you'd benefit from the better station. Currently, the FX888 type technology is the lowest cost that still meets IPC standards.
However, as production methods improve, the maker may find out that its no longer cost effective to manufacture low cost stations, then you are forced to move up, just like with DSOs, the 25 MHz models cost almost as much as the 50 MHz, at least comparing the Rigol 1052e to other brands. But until that day comes, or parts are not available, a buyer needs to see what is the most cost effective for their level of work.