Performance improvement - handling buffers in plp.rs

[JasonCookSyncadd]

Requesting single bytes from a tcp/ip stack incurs significant overhead.

I made a quick change to buffer coercion in src/tds/codec/column_data/plp.rs and in my specific case, performance improved by a factor of 3 (in my case when utilizing varchar).

The changes I made were as follows: in fn decode (comment this for loop out) //for _ in 0..len { // data.push(src.read_u8().await?); //} (replace with this) data.resize(len as usize, 0u8); // actually allocate the space src.read(&mut data).await?; // bulk read the data

Last else loop of decode (my case did not test this, concept code only): //let byte = src.read_u8().await?; //chunk_data_left -= 1;

//data.push(byte); let mut buffer=Vec::with_capacity(chunk_data_left); // create a temporary storage for block buffer.resize(chunk_data_left,0u8); // actually allocate the space src.read(&mut buffer).await?; // bulk read the data data.extend_from_slice(&buffer); chunk_data_left=0;