In any plan with dependencies, one chain of tasks determines the end date. That is the critical path, and it has zero slack: any delay on it moves the finish.
Everything off it has float. A task with five days of float can slip four and change nothing, which is why treating every delay as equally urgent wastes attention.
When the date cannot move, float is the only resource you have. Knowing where it is tells you where to push and where to leave people alone.
Without it, a two-day supplier delay produces either panic or indifference, and which one you get depends on who noticed rather than on whether it mattered.
If there are real dependencies and a fixed date, yes. Otherwise a task list is fine.
Recording preference as dependency, which invents a critical path and hides real float.
Constantly. Calculated once, it is a historical document.
Half an hour with your own data usually saves reading three of these. The guides will still be here afterwards.