ESU LokSound Select Micro Install in Kato GS-4

[Mark W]

...The LED is a diode and will block the back side...

True to some extent, but an LED is in no way as reliable or SAFE as a proper diode at blocking current the wrong way.
A simple google search yields a heavy majority of sources cautioning against such use.

Yes, our experiences are different. Yet I still will prefer to use one less wire when possible.

I would agree one less wire is a benefit, but not at the risk of an unreliable and unsafe method.
Would it work most of the time either way?  Probably.  But anecdotes are not scientific, physics of proper diodes are. 

[woodone]

I have been doing DCC installs for over 15 years, to date I have yet to see a single failure using this method.
It is true that not all of my installs have used this method. But with the OP loco I have done over 200 Kato N scale GS-4 installs and most used 3 LED's and NO blue wire connections.
I don't see this has a unreliable nor a unsafe method.
But that is not just MO, but that fact that it has worked.

[peteski]

Why add a circuit when deleting a wire will do the same thing?

It seems to me that you are missing the point. Let's go over the scenario.  Steam loco. Decoder in the tender along with electric pickup from the wheels. The loco can also pick up electricity from the track, has a motor and a headlight.

The goal here is to delete the blue wire connecting the tender with the loco. In your case you are either using the existing connection from the locos headlight to one of the rail pickups or connecting a new wire between the headlight and the rail pickup.

In the solution I presented, the blue wire connection between the tender and loco is also eliminated, and all you have to do (in the loco) is to install tiny dual-diode components and run 1 wire from the headlight to the added component and then 2 more wires between the component and the power pickups. It really isn't very complicated or take up much room.  The added bonus is that the headlight now runs from DC (not pulsing DCC track voltage) making it fully compatible and functional under all possible DCC operating modes (including zero-stretching) and with DC operation too.  IMO, well worth adding the diodes and 2 more wires than your example.  But I'm not trying to persuade anybody to follow my example - just giving the option to make the model fully compatible with all the possible modes of operation.

There is no need to keep arguing merits of each method - yours works, while the one I presented (but not invented) has a wider compatibility range.