conv2

chore: Release conv2 version 0.4.2

fix warnings in bench

use a simpler float->int impl for large integers

add a bunch of new unit tests

add new impl for float2int truncate modes