Global moves with proposal probability on log scale

[johanneshofmann87]

In some cases, the proposal probability can be extreamly small or large, reaching the range of double precision. This will cause numerical issues. Hence, the goal is to provide a new subroutine where the T0_proposal_ratio will be on a logarithmic scale. The base implementation can be a wrapper of the current Global_move function to maintain backwards compatibility.