The trem speed pot can go to the star ground (because the trem circuit gets it’s power supply from the smoothing (screen) cap supply node (and the trem functions via the output tubes). The intensity pot can stay where it is, as long as the bias supply is also grounded with the other ‘high current’ returns. (The bias supply works through the output tubes).
The reservoir and smoothing cap nodes (The main Pii filter) i.e. the cap can ground, should also ground at the same place. The heater ground reference (or heater elevation circuit ground) should also ground here, along with the High Tension winding CT, and the 6V6 cathodes (or the bias measuring resistors that are in series with the 6V6 cathodes - if you are using these).
Then, if you need/want to go to Merlin’s galactic ground later on, it’s a simple matter of moving this cluster to the floating end of the ground buss wire that I suggested earlier. And this is what I was talking about earlier when I said that ideally the amp only wants one signal ground return to the chassis if minimising ground hum is the goal. But the reason that you’d do it this way (I.e. a galactic ground instead of a star ground point), is you get better control of the sequence in which the high-current vs low-current returns are ‘stacked’ together- with the low current returns being immediately at the chassis (at the input jack), without the high-current returns being in between.
But see how you go with the split ground first. Split grounds work well in lots of amps. (The only real reason for keeping a galactic ground option open in this case, is the noise-sensitive reverb recovery stage, and the star ground is potentially problematic in this case).