43 servo motors through transistors, jittering

A gross oversimplification. Find the spec sheet for YOUR SG90. Here's a snippet from one that is available from Amazon:
Typical hobby servo

Relevant specs from that page:
Running current (at no load): 400±30mA (4.8V ) 500±30mA(6V)
Stall current (at locked): 1300±40mA (4.8V ) 1600±50mA(6v)
Idle current (at stopped): 6±1mA (4.8V ) 6±1mA(6v)

So, you need to do a bit of digging. Find your spec sheet. Know your application. How many servos will be commanded to move at any one time; how many may be in stall, unable to reach end-of-travel; how many may be running at any one time?
The following is based on the numbers for that servo. Consider, a servo at rest which receives a change in position draws the stall current until it actually starts moving. So if you command your 43 servos to change position (you wired them all from one output, right?), the instantaneous demand is something like 50+ amps! Once they are moving, you're actually looking at 43*400 mA, so 17.2 Amperes running current until they reach new setpoint. Sure, it won't be a long demand, but it does exist. Better you should stagger your position changes to avoid this.
That's a start. If any/all of your servos can actually fail to reach commanded position due to mechanical interference, it gets worse, because those servos will draw 1.3 A until you turn them off, or they burn out. 1.3A at 5V is similar to the old glass 7W Christmas tree bulbs - they got plenty hot!
C