Are we seeing the full benefits of LTO (link-time optimization) in CFU Playground?

With the picolibc integration in LiteX, link-time optimization is possible. But I'm not sure if we are seeing the benefits in *all* of the software that we compile into our monolithic executable, which includes not just LiteX libraries but also the TFLM library as well as our own code (menus etc.). Performance-wise, the TFLM libraries would be key.

@kgugala , I think you reported a 20% boost from initially enabling picolibc; but as we have seen elsewhere (#324), simply changing the code layout can increase/decrease performance by +/-20% or more, so it may be difficult to analyze.