Possible incorrect calculated start date when toggling duration mode of given dep-free task

[dylms]

When toggling duration mode of a task, without changing its actual period (same start and end dates), the task below (if it doesn't have any dependency nor independant start date, then relying to pre-#35-fix behavior and induct dep to previous task) can have a bad recalculated start date.

[dylms]

Delta recalculated in model.py:process_gantt() after toggling to duration mode (erasing previous task end date) doesn't take in account closing days and possible resource leaves which lengthens previous task and so below task's start date.

Fix : if below task is in duration mode without dep, set its start to gantt calculated end of previous task instead of calculating a delta only based on prev' task duration.

Another problem was that LeaveDatas containing leaves periods of resources are processed AFTER TaskDatas, so resources of the task are considered present during all the latter (because vacation property has not yet populated) when calculating prev' task end date. Interverting AbstractTaskData and LeaveData lists in the __list variable of the planning, hence process leaves BEFORE tasks permits to take them in account when calculating end date.