Rename CanvasItem.update() to queue_redraw()

[Mickeon]

Closes godotengine/godot-proposals#5190.

As brought up in #54161 (comment), #54161 (comment) and Should CanvasItem.update() be renamed to request_redraw() or redraw().

Affects a lot of classes. This PR is absolutely, utterly GARGANTUAN.

I very thoroughly checked signal connections and deferred calls to this method, add_do_method/add_undo_method calls, and so on. Just a few comments have also been changed to say "redraw". It's ENTIRELY possible I still missed something, but the engine compiles, at least!

Also renames the internal _update_callback() to _redraw_callback() for consistency.

In CPUParticles2D, there was a private variable with the same name. For now, this has been renamed to do_redraw. It's either that, or something close. I haven't exactly understood what it's meant for.

This PR was previously renaming it to redraw()

[YuriSizov]

We've discussed this in a bikeshedding meeting, and were convinced that a rename would be beneficial here. But we suggest using queue_redraw as a name. It's probably worth renaming the internal function, _update_callback, to make it clear it's only for redrawing and not any other kind of updating.

[Mickeon]

There are occasions where update() is actually called externally by a parent. For example:

godot/editor/plugins/tiles/tile_data_editors.cpp

Lines 701 to 718 in dc4193b

	void GenericTilePolygonEditor::set_polygon(int p_polygon_index, Vector<Point2> p_polygon) {
	ERR_FAIL_INDEX(p_polygon_index, (int)polygons.size());
	ERR_FAIL_COND(p_polygon.size() < 3);
	polygons[p_polygon_index] = p_polygon;
	button_edit->set_pressed(true);
	base_control->update();
	}
	Vector<Point2> GenericTilePolygonEditor::get_polygon(int p_polygon_index) {
	ERR_FAIL_INDEX_V(p_polygon_index, (int)polygons.size(), Vector<Point2>());
	return polygons[p_polygon_index];
	}
	void GenericTilePolygonEditor::set_polygons_color(Color p_color) {
	polygon_color = p_color;
	base_control->update();
	}

By the naming conventions I've seen, is it still a good idea to rename the internal method _update_callback?

[YuriSizov]

There are occasions where update() is actually called externally by a parent.

Yes? That's pretty normal even for user scripts.

By the naming conventions I've seen, is it still a good idea to rename the internal method _update_callback?

What do you mean? We're renaming the public method to make it clearer what it does, so the internal one needs to be renamed as well to avoid any confusion for engine developers. The name would be something _redraw_callback, so I don't see any problem with that.

[Mickeon]

Oh I see! I assumed the starting underscore was mostly reserved for virtual or private methods. I am a little confused on _update_callback

[YuriSizov]

Oh I see! I assumed the starting underscore was mostly reserved for virtual or private methods.

_update_callback is and _redraw_callback should still remain a private method.

Again, public method is changed from update to queue_redraw. Its internal counterpart that is actually called to perform the redrawing is changed from _update_callback to _redraw_callback. Two methods are getting renamed.

[Mickeon]

Oooh I see, I understand it all now. Thank you very much

[Mickeon]

Updated the PR to use queue_redraw. To be honest, I do still prefer redraw over it, because for being so commonly utilised in the codebase it may be bit annoying to see, and the documentation of this method I updated goes into greater detail on how it works, but I've seen a few methods around with similar names and they do actually redraw on the spot, so... I suppose it's not worth splitting hair for.

[YuriSizov]

I am tentatively ready to approve this, but it needs another rebase and quick look in case you accidentally renamed stuff you shouldn't have or missed stuff that you should've renamed (though unlikely if you've only did the bulk of changes in various cpp files).

Edit: Also, you didn't rename _update_callback to _redraw_callback?

[YuriSizov]

Updated the PR to use queue_redraw. To be honest, I do still prefer redraw over it, because for being so commonly utilised in the codebase it may be bit annoying to see, and the documentation of this method I updated goes into greater detail on how it works, but I've seen a few methods around with similar names and they do actually redraw on the spot, so... I suppose it's not worth splitting hair for.

I think it's important to indicate that the method has no immediate effect. We use queue_* for other method names as well, including widely utilized queue_free(), so it's not a foreign concept.

[Mickeon]

Edit: Also, you didn't rename _update_callback to _redraw_callback?

I first forgot, then I thought of doing it in another PR but it doesn't hurt to just do it now at this point. Won't take too much.

[Mickeon]

All done, I had checked one more time.

[YuriSizov]

Changeset seems good now. Didn't check every single line, but in principle it should be correct if the CI passes

[YuriSizov]