Ran your data through GRT, using the default inputs where not available from you - the most useful/helpful data point would be knowing the average "H2O capacity" of a sample of fired cartridge cases to get a more accurate simulation of your load.
That being said, the initial output from GRT gives a predicted MV of 2400 ft/s - pretty darn close to your measured MV. The peak pressure for this load is calculated at 36000 psi - which is very low and is a likely source of wacky/inconsistent performance. At such low pressure the neck seal in the chamber on firing will be poor (giving very dirty case necks on the outside).
Interestingly, your 44.0gr load theoretically should be very close to an accuracy node.
Your next node up should be around 45.4gr for an MV of 2460ft/s (node 7)
Node 6 is estimated at 48.1gr for an MV of 2612 ft/s and a peak pressure of 49k psi. This is getting closer to the ideal operating conditions IMO.
If possible load up some test loads in +0.5gr increments and give us the MV's and we can finetune from there. The usual caveats apply - the ONLY change from the data you supplied must be the load increments.
I think you'll run out of case capacity before you hit any pressure limits but if/once you hit 2700ft/s, please stop cos you'll be reaching Pmax, aka the red zone. Pmax is only 56500 psi in the old x57, so not too demanding in a modern action - so I wouldn't be too tentative about pushing a load close to this point.
A side note - RL-19 is not really optimal, it's too slow burning. It's likely that you'll have a lot of soot on case necks, in the chamber and a fair amount of muzzle flash from powder burning outside the muzzle.
H4350 would be better, although still not 100% burn - RL-17 would be ideal for this combo