Buffered IO is about 3-4x slower than emp netio

[xiangxiecrypto]

Need to check out the implementation details.

[xiangxiecrypto]

emp c++ performance:

emp rust performance:

[ujnss]

The main reason is that the c++ version uses the memory address of the block array directly, while the rust version sends or receives blocks one by one.

Optimization. When sending, convert block arrays to [u8] first. When receiving, receive [u8] and then convert to block array. See #11.

[xiangxiecrypto]

improved performance in emp-rust