Ordering issue in risk disaggregation

[Michele Simionato]

Theew is an ugly ordering problem in the code for risk disaggregation. In particular the rupture_ids are extracted from out.output.event_loss_table.keys() in an unspecified order. This as issue because later in the code there is a line zip(loss_ratios, rupture_ids) which assumes that the loss_ratios are in a fixed order. So in practice the associations rupture <-> loss_ratio are all wrong and we cannot trust the disaggregation results for loss fractions.

[Michele Simionato (michele-simionato)]

https://github.com/gem/oq-engine/pull/1491