I see. Grid can be used like that, but it’s not the recommended or common workflow. Typically you select a grid which is fine pitched enough to represent all used coordinates and then just translate the datasheet values to coordinates in one way or another and draw items. Editing items happens WYSIWYG by grabbing and dragging their points afterwards, or editing their property values.
This is not very difficult, everyone does it like that and gets used to it. There are some “invisible” features to help editing, like hitting spacebar to set the “local origin” and see relative coordinates in the bottom of the view, and using shift, alt and control keys while drawing (I don’t remember right now what does what, they have been changed for the unstable development version 5.99 which I have used more recently).
As for constraints, they would be really useful; see Constraint based component placement.