Will indeed not make a big difference and probably be even slower. It is more like a leftover from an experiment (should be cleaned when I find the time). In B4J you simply use logic numeric values for the row and cell (e.g. Cell(1,2)), but internally they can be quite complex IDs (holding parents, arrays, etc) because in the HTML they need to be unique.
The container holds a map with the complex row and cell each component is located in.
e.g. key: compA, value: a pair object with Row = complexRowID, Cell = complexCellID
So if you do Component("compA"):
1. it get the pair object for compA first
2. gets the real row using complexRowID
3. get the real cell using complexCellID
4. gets the component
if you do ComponentRowCell(y,x):
1. build the complexRowID/complexCellID
2. gets the real row using complexRowID
3. get the real cell using complexCellID
4. gets the component
I always use the Component() method.