Improving the perfomance of the coro-prime-sieve algorithm in C# implementation

[anc-u]

I kept the algorithm as it was, but removed the memory allocation for the utf16 string for the number when writing to the console, and also added synchronous processing where you don't need to wait to write to the channel from another thread, because in C# the channel interface allows you to do this