The web api of vMix is slow. Happy that you just have 200ms of delay.

The best is the 4. You just have to remove the input at the end of his play back:

1- The first clip will have obviously a delay
2- Is it a realtime solution i.e the media has to be played as soon as the user or the rundown request it or can he choose to postpone as he want.
3- with your app, fetch the duration of the clip (there is third party dll that can help here or you can do it with vMixAPI later). if the playback don't start at the beginning and don't end at the end of the clip, you do simple math (end-begin)
3- when the user or your rundown request to play a clip, create first unique guid with your app then send addinput command to vMix with that guid(see documentation)
4- If it's realtime transition/set to output/overlay/play (depending of how you want that clip to play) into that input providing the input guid or wait the starting time choosen by the user then transition.
5- if you didn't manage to get the duration of the clip, send an empty request to vMix and find your guid in the list of input. there is a duration parameter.
6- Do the same if you want to add another clip after this. Just start the process sometime before the end of the previous(i.e 1000ms) and then transition as soon as the other finished (remember you have store the duration to help you for that) or or if you want to play it directly you can transition over the previous input
7- Wait some seconds and remove the previous input (you have his guid) via the webapi.
8- and so on...

You can use only 2 unique guid. when you remove one input from vMix interface you resuse his guid for upcoming clip.
If your app is well done, you are good to go. This is how the vScheduler work and I think it's difficult to do it another way. You can check out the source code for more ideas!

No socket api. Only the Web Api. It use Web request the one you can use in any web browser not TCP/IP.

For your problem there are solutions.
Add shortcuts features to your automation app (I don't known how your automation work so)
You can program the shortcuts in you automation app for sending command concerning input sent by your app. And don't use those shortcuts in vMix. When the user press a specified key, the app send what ever command you want to the vMix with the right input guid since you have that information. You can add in The UI of you app a place where the user can choose key and shortcut.

You can build another app that will handle the shortcuts.
You use always the same 2 guid in you automation app. When you remove one input, reuse the guid of that input the next time you need to create a new input. So the new app just need to send what ever command is assigned to a keypressed to one of those 2 guid. If one failed try to send to the second.

You can uncheck the option "assign shortcuts by input" in vMix if you want vMix to use guid.
- Let you automation app create input;
- go in shortcuts menu and select the input created by your automation.
- Duplicate that shortcut and wait your automation to create the second input, then assign the duplicated shortcut to that second input.
- Now you will have duplicated shortcut for the two guid. Don't worry about duplicate it will works.
OR
- Just create shorcuts and assign them to input 1 and duplicate assign the duplicate to input2. Uncheck assign by input, then save.
- Go in vMix install folder, in shortcuts folders you will see a file corresponding to you shortcuts config. Open it. It's a XML document. There is a field named key for every shortcut. Just paste the first guid you use in those filed for the first set of shortcut and the second guid for the duplicated one.

The downside of those solution is that for a few time before your remove the previous input after transition you will have your 2 guid available. Just wait 1-2 seconds for the automation to remove completed input before using shortcuts. In the first two solutions you can check the playtime of input to know which of the two you need to send command.