[Feature Request] Performance test automation

[lsylsy2]

Creating this issue as a feature request item for myself, planning to contribute and also use to close #327
The automation should:

1. Run on a physical machine (Linux? Including rpi or similar boards for ARM / routers) or VM with stable performance.
2. Run different versions of UDPspeeder, run performace test using iperf or similar tool, measure the detailed CPU usage / bandwidth / etc.
3. Use netns / tc / etc. to simulate latency and packet drop

[lsylsy2]

In #326 there was discussion:

Also multiple-core is not the biggest possible improvement. I actually have investigated a avx2 enabled FEC library which can improve the performance by 10x，even if with single-core.   (Combining avx2 and multiple core, it's not hard to achieve a performance improvement of 40x)

However, avx2 optimization is not suitable for personal use; what's more important is to improve performance / efficiency on low-end devices like low-end routers, battery-powered phones, cheap virtual servers, thus make the project usable to more people, instead of empowering someone to use Gigabits of bandwidth with a high-end server.