Add support for precision truncation of QPU problem data coefficients

[randomir]

When used together with QPU problem data compression on upload, upload size is reduced significantly (compression ratio increases because of all the zeros in problem data).