The RPi GPIO pins support the right voltage but each can supply only a maximum of about 16mA, so it would be dangerous of me to try connecting the laser's +V line directly to a GPIO pin.
Yes, you can't supply the laser directly from a GPIO pin.
I'd have to incorporate a voltage divider to ensure it supplies 3V
I don't think a voltage divider is a good solution, I suggest using a voltage regulator. Check LM317.
As @goldilocks suggested, you can use the 3.3V rail since it can supply up to 50mA. You can use a 10ohm resistor (8.2ohm would be better) in series with the laser to drop the excess 0.3V.
This post recommends I buffer the GPIO pin an NPN transistor, with current coming from the RPi's 5V rail
Yes, thats the way to go, but instead of the 5V rail, you will have to connect to the output of the regulator 3.3V rail in series with the resistor suggested above. The Vce of the BJT transistor would be around 0.2V, so you wouldn't really need a resistor since the voltage around the laser would be 3.1V.