Update readme with suggestion on number of threads to use

[HarryGifford]

Update the readme to make comment on performance with hyperthreading and link to pytorch docs which have relevant information.

For example, testing on my laptop (AMD 6850U 8 cores) I saw a 3x speedup running stories42M.bin by going from 16 threads (default) to 8 threads:

(mypyenv) hegiarch% ./run ./stories42M.bin -s 42                  
Once upon a time, there was a little girl named Lily. She loved to play in the park with her friends. One day, she found a shiny green gem on the ground. She picked it up and showed it to her friend, Max.
Max said, "Wow, that's a cool gem! Can I see it?"
Lily replied, "Sure, but be careful not to break it."
Max looked at the gem and said, "I wish I could find a gem like that in real life."
Lily said, "I know, but I don't want to lose it. I value it very much."
Max agreed and they continued to play in the park. But Lily couldn't stop thinking about the green gem. She decided to keep it safe in her pocket. She knew it was special and she wanted to cherish it forever.
achieved tok/s: 48.049600

(mypyenv) hegiarch% OMP_NUM_THREADS=8 ./run ./stories42M.bin -s 42
Once upon a time, there was a little girl named Lily. She loved to play in the park with her friends. One day, she found a shiny green gem on the ground. She picked it up and showed it to her friend, Max.
Max said, "Wow, that's a cool gem! Can I see it?"
Lily replied, "Sure, but be careful not to break it."
Max looked at the gem and said, "I wish I could find a gem like that in real life."
Lily said, "I know, but I don't want to lose it. I value it very much."
Max agreed and they continued to play in the park. But Lily couldn't stop thinking about the green gem. She decided to keep it safe in her pocket. She knew it was special and she wanted to cherish it forever.
achieved tok/s: 186.746988