Designing Nucleotide Sequences for Computation: A Survey of Constraints