1 kHz BW limit for the TRMS mode is a bit low. However it depends on how sharp the limit is. A lower BW can also have the positive side of seeing less noise.
For digital RMS one does not necessary need a lot of oversampling. Just some pseudo random sampling timing is enough and could even work with undersampling (e.g. 1 MHz BW with 100 kHz sampling).
The BW limit may as well be due to the analog input part.
The BW specs for the analog RMS chips are also a bit tricky: at lower amplitude the BW can go down and this can cause linearity errors. At some 5 or 10% of the fullscale (this may be a range one may have to chose for some signals) the BW may no longer be that large.
It looks funny to have a maximum speed of some 65 readings per second - that is just a little faster than 1 PLC. So one would get only slightly higher speed at the price of no longer mains hum supression and possible aliasing that can cause confusion. I would normally consider the range of 51/61 SPS to some 250 SPS as relatively useless, as it is effected by mains hum, but can not really resolve it. If the meter does RMS the digital way, which is a reasonable decision, they should be able to do faster readings (e.g. 2 kSPS) too. Maybe they should switch from slow RS232 to USB.