Add level 2 compiler optimizations

[Loomeh]

Thanks to pull request #18, LAST now seems to work with level 2 compiler optimizations. I tested a few levels of Jet Set Radio and didn't seem to get any segfaults or other issues.

I would like this to be tested across different distros and hardware before merging though.

My setup for comparison:

OS: Arch Linux
GPU: NVIDIA GeForce RTX 3050 Mobile
CPU: Intel Core i5-11400H

[EXtremeExploit]

works on my machine:

OS: Arch Linux (6.7.9-zen1-1-zen)
GPU: ASRock Phantom RX 580 8GB Gaming D
CPU: AMD Ryzen 2600
make version: 4.4.1
gcc version: 13.2.1 20230801

Tested with https://github.com/EXtremeExploit/mlp-amba-autosplits/blob/master/MLP%20AMBA.lua

[wins1ey]

Works on my machine:

OS: Arch Linux (6.7.9 kernel)
GPU: Nvidia 3060 Ti
CPU: Intel i7-5775C
make version: 4.4.1
gcc version: 13.2.1