Track peak memory usage

[PyXiion]

From #10

While running benchmarks, I encountered an unexpected and massive spike in memory consumption. Some libraries hit nearly 28 GiB of RAM, bringing my system almost to death. Luckily, I managed to kill the process just in time.

I think memory usage tracking will be very useful.

Additionally, it would be helpful to add a warning in the README to alert users about potential high memory usage.

Roughly what the table will look like

Runtime	libfork	TooManyCooks	TooManyCooks_st_asio	TooManyCooks_mt	libcoro	libcoro_mt	cobalt_st_asio	cobalt
skynet	15.97 MB	15.97 MB	N/A	N/A	20.62 GB	N/A	N/A	N/A
nqueens	15.97 MB	15.97 MB	N/A	N/A	5.37 GB	N/A	N/A	N/A
fib(39)	15.97 MB	15.97 MB	N/A	N/A	17.02 GB	N/A	N/A	N/A
matmul(2048)	59.66 MB	57.09 MB	N/A	N/A	60.93 MB	N/A	N/A	N/A
io_socket_st	N/A	15.97 MB	N/A	N/A	15.97 MB	N/A	N/A	16.1 MB
channel	N/A	N/A	28.76 MB	31.39 MB	N/A	15.97 MB	15.54 MB	N/A

[tzcnt]

I see this is still in draft status. I was able to run and get memory statistics on Linux (which I'll update the README with). So I think it's pretty much ready. Thanks for showing me about these OS APIs, it's very helpful to know.

On Windows I had to swap the order in memusage.hpp so #include <windows.h> comes before #include <psapi.h> so if you don't mind swapping that I'd be happy to merge this. There are some other unrelated build issues on Windows but those are on me to resolve.

[PyXiion]

I want to make memory stats in the graph too, but I think I'll implement this a bit later.

I have no Windows machine, so I wrote implementation based on docs. I think CI wouldn't harm to add later