[1.19] Improve duplicate checks for implicit counts of 1

[rlnt]

The sanitizing was initially applied to all possible elements but this obviously drastically increased load times. I did some benchmarking on the dev version of Enigmatica 9 with EE. Here are the results.

Old version (no implicit count checks): 8.853ms

New version (sanitized every element): 18.330ms

With basic type filter: 17.273ms

if (firstElem.getClass() == secondElem.getClass() ||
    (firstElem instanceof JsonPrimitive && secondElem instanceof JsonObject) ||
    (firstElem instanceof JsonObject && secondElem instanceof JsonPrimitive)) {
    firstElem = sanitize(firstElem);
    secondElem = sanitize(secondElem);
}

With basic type filter and array length checks: 17.068ms

if ((firstElem instanceof JsonArray firstArray && secondElem instanceof JsonArray secondArray &&
     firstArray.size() == secondArray.size()) ||
    firstElem.getClass() == secondElem.getClass() ||
    (firstElem instanceof JsonPrimitive && secondElem instanceof JsonObject) ||
    (firstElem instanceof JsonObject && secondElem instanceof JsonPrimitive)) {
    firstElem = sanitize(firstElem);
    secondElem = sanitize(secondElem);
}

With type filter, array length checks and without primitives: 14.593ms

if ((firstElem instanceof JsonArray firstArray && secondElem instanceof JsonArray secondArray &&
     firstArray.size() == secondArray.size()) ||
    (firstElem instanceof JsonObject && secondElem instanceof JsonObject) ||
    (firstElem instanceof JsonPrimitive && secondElem instanceof JsonObject) ||
    (firstElem instanceof JsonObject && secondElem instanceof JsonPrimitive)) {
    firstElem = sanitize(firstElem);
    secondElem = sanitize(secondElem);
}

The last method has an insanely ugly condition but it's only 5.740ms slower than the version without any sanitizing and implicit count checks at all.