fix: optimize the algorithm for cpu count calculation in the linux

[beita1]

Background

• This pr is created as a result of discussion. We found the max thread count when equal to the number of physical cores is the best way, but not before.

Content

• We use the command cat /proc/cpuinfo | grep "physical id" | sort |uniq | wc -l to get the physical CPU number and use the command cat /proc/cpuinfo | grep "core id" | sort | uniq | wc -l to get physical CPU core number in each CPU. Finally, we can get the physical CPU cores number by multiplying these two values together.

[Aslemammad]

Thank you so much for teaching me and sending a PR :)