"low latency" automation/routing would be digital like RME total mix and so has slight latency.

The lowest latency (as close to true zero as possible) would be analogue with flying fakers or VCA. very rare.

with flying fader and VCA you can automate levels and still have "zero latency" (close to the speed of light A>B???)
for your signal to pass through your gear.

Sorry, but I do not get your point here, please explain what you mean.
Describe the audio flow from source to destination that you mean, and what concretely worries you where.

So your setup is this?

PC
|
|USB
|
UCX II------ADAT IN<-------[opt. SPDIF]--------SPDIF OUT---some other device
        \
         +--SPDIF OUT--------[coax. SPDIF]------->SPDIF IN---external DAC---

|-----lossless transport of digit. Audio data---------------------| |-----------|
                                                                                             ^------ D/A conversion on external DAC (the "one and only")

Audio from, e.g., a player on the PC is digitally / lossless being transported from the player through USB, through coax SPDIF up to the external DAC which makes the one and only D/A conversion. So where is your point / issue?

If you had an ADI-2 DAC or Pro FS as "DAC", then you could validate lossless audio transport yourself using the Bittest, see manual or this diagram of one of my blog articles here:

In the driver settings "WDM submenu" you can create Windows compatible device entries for every pair of input and output port. One of the output ports you can define as a speaker device. Then you find it easier in the Windows audio settings as "Speaker device" which you can use as the default Windows audio device. This is for Windows and all applications which do not support ASIO, like an internet browser for playing YouTube stuff, etc.
In the Windows audio settings, you can right-click the device and make it as your default playback or communication device.

For the rest of your question. Simply select the input or output port to create a WDM device, which you need on Windows or for Windows application without ASIO Support. You should know what's connected where … Select those ports, where you require it.

The speaker device is additionally there because in the WDM settings you selected the SPDIF coax output as speaker device, then you create an additional "speaker device" which you can identify it easier. There should be no functional difference.

It doesn't matter to which output you send audio ..  The audio is not being sent to the real HW output, instead of this it 1st shows up under SW playbacks. This is any audio coming from the PC.

The final routing decides which audio from HW inputs (top row) and SW playbacks (middle row) you want to hear.
Each of the HW outputs has a submix of its own.

So the operation in submix is "isi pisi" this way.
You select an HW output.
Once you did that you see the routing / submix for this particular HW output.

In your case select the HW output for the SPDIF coax output.
Now you see the submix for this particular output.
Watch the fader positions in the top and middle row. This is your submix.

Do you send in foobar audio to AN1/2, then you see audio arriving in the middle row at AN1/2.
Raise the fader of SW Playback AN1/2 and you hear audio on your SPDIF coax port.

Did yo send audio in foobar to AN3/4 .. well same for AN3/4 ...
Raise the fader of SW playback AN 3/4.

So .. this way you can organize your audio playback on PC. Use the middle row like a patchbay.
Send OS sound to (SW Playback) AN 1/2, MusicBee Player to 3/4, Audio from Games to 5/6, the DAW Sum to 7/8 ...

Now you can use the flexibility of TM FX to create for every HW output in the bottom row the ideal submix of inputs and software playbacks ....