[vms/avm] Replace strings.Replace with fmt.Sprintf in tests

[dhrubabasu]

Why this should be merged

@ARR4N pointed out that we could use fmt.Sprintf with explicit arg indices for better readability during review of another PR (#3165 (comment))

How this works

Replaces strings.Replace with fmt.Sprintf in the avm tests (only place we do this pattern currently)

How this was tested

CI

[ARR4N]

Approving with a requested change that doesn't need re-review.

[tsachiherman]

I think it's much better with these changes; next step, I'd suggest moving these large json structure into an external file and embed these doing buildtime. ( I wouldn't make this suggestion for few lines of json.. but these jsons are quite massive, and look like they might be changing regardless of the testing code ).

[marun]

Assuming use of json.Marshal/Unmarshal in the code under test, would it make sense to switch to composing JSON test data in the same manner rather than via string interpolation?

[dhrubabasu]

I think it's much better with these changes; next step, I'd suggest moving these large json structure into an external file and embed these doing buildtime. ( I wouldn't make this suggestion for few lines of json.. but these jsons are quite massive, and look like they might be changing regardless of the testing code ).

These won't be changing at all. The reason the JSON is in the test code is to guarantee we don't make a change that accidentally changes a JSON serialization (which would impact downstream consumers, particularly in javascript land)

[dhrubabasu]

Assuming use of json.Marshal/Unmarshal in the code under test, would it make sense to switch to composing JSON test data in the same manner rather than via string interpolation?

I like the idea of keeping the JSON itself since Marshal versus Unmarshal could lead to an undesired change (serializing null instead of [] for example)