User interaction can be a valid part of a program
There is an existing consensus about programming languages which require user interaction to run, specifically that they are valid languages with valid answers despite said interaction. The question then becomes, how does that apply to individual Programs which require user interaction to run when their language doesn't universally require it. I would roughly break it down into two categories:
User Interaction defined in the source code
The first category is Languages which represent user interaction in the source code. In the linked consensus, Vim is like this; "programs" written in Vim represent a series of keys that a user will press to execute the program, with each user interaction accounted for and represented in the byte count. I would apply this to a hypothetical mixed language that can include user interaction, as long as that user interaction is strictly defined within the source code and can be translated to bytes. If the answer in question were written in a language which defined a symbol that means "user presses enter" and otherwise behaved identical to Command Prompt, then the answer would be valid with that 1 symbol added (though doing so now would likely fall afoul of a forbidden loophole).
User Interaction standardized for a language
The second category is Languages which require user "cranking" to execute. These languages in their base definition do not define a way to advance stages of execution, and substitute a standardized way for the user to run them which includes a pre-defined interaction. This could also be applied to a mixed language, such as one which can do normal sequential execution but requires a held enter key for looping. The answer could fit here, and then it would be valid with 0 additional bytes. However, as in this example, Command Prompt (and related "Batch") does have looping structures which don't require user input. Using a held enter key to cause looping behavior is not a standard of the language, it is extra user interaction that the author tacked on to the answer.
User interaction can also be invalid
In my opinion, the linked answer does not fit in either of the above categories that would allow user interaction to be specified within source code (thus adding to byte count), or have user interaction as a standard of the language and thus implied by the language (since knowledge of the language used is treated as "free"). In this case, having a User hold the enter key (or keep it pressed some other way) would instead count as an "additional input" that indicates the duration that the program should continue; even if the remaining text of the post indicates that said input should have a constant value ("forever") it still violates the loophole forbidding additional inputs and thus is invalid (specifically a Type 2 invalid answer).