Fix WebSocket threads CPU usage

[afondard]

What does this PR do ?

This PR optimizes the WebSocket reconnection logic in the Kuzzle SDK to address high CPU usage during frequent reconnect attempts.

How should this be manually tested?

• Step 1: Start the Kuzzle backend.
• Step 2: Write a test Java application using the updated SDK. You can build the jar and then add the dependency in your configuration file like described below.

Maven: Add the updated SDK jar to your project’s root, then add the following dependency to your pom.xml:

        <!-- Kuzzle JVM SDK -->
        <dependency>
            <groupId>io.kuzzle</groupId>
            <artifactId>sdk-jvm</artifactId>
            <version>1.3.3</version>
            <scope>system</scope>
            <systemPath>${project.basedir}/sdk-jvm-1.3.3.jar</systemPath>
        </dependency>

Gradle: Add the updated SDK jar to your project’s lib/ folder and update your build.gradle as follows:

        // Define repository for local jars
        repositories {
            flatDir {
                dirs 'libs'
            }
        }
        
        dependencies {
            implementation name: 'sdk-jvm-1.3.3'
        }

Here's an example of a test Java main:

package kuzzle.kuzzleio;

import io.kuzzle.sdk.Kuzzle;
import io.kuzzle.sdk.coreClasses.responses.Response;
import io.kuzzle.sdk.protocol.WebSocket;

import java.util.HashMap;
import java.util.Map;
import java.util.concurrent.ExecutionException;
import java.util.Scanner;

public class Main {
    public static void main(String[] args) throws InterruptedException, ExecutionException {
        System.out.println("Hello world!");
        WebSocket ws = new WebSocket("localhost", 7512, false, true, 1000L);
        var kuzzle = new Kuzzle(ws);
        kuzzle.connect();
        System.out.println("Connected to Kuzzle server successfully!");

        Map<String, Object> query = new HashMap<>();
        query.put("controller", "document");
        query.put("action", "get");
        query.put("index", "tenant-asset_tracking-toto");
        query.put("collection", "devices");
        query.put("_id", "Tool-Tool1");
        Response res = kuzzle.query(query).get();
        System.out.println(res);

        Scanner scanner = new Scanner(System.in);
        System.out.println("Please enter your first name:");
        // Arrêter le back ici pour tester la queue
        String firstName = scanner.nextLine();
        System.out.println("Hello " + firstName);

        Response res2 = kuzzle.query(query).get();
        System.out.println(res2);
        // Relancer le back pour se reco et faire la requête
        System.out.println("Please enter your last name:");
        String lastName = scanner.nextLine();
        System.out.println("Last name: " + lastName);
        scanner.close();

        Response res3 = kuzzle.query(query).get();
        System.out.println(res3);
    }
}

• Step 3 : Run the test Java application and simulate network interruptions while monitoring the CPU usage (top command for Linux).

[alexandrebouthinon]

This PR break the async behaviour of the Java SDK when using the WebSocket protocol. You must revert your changes to avoid hidden breaking change on runtime. By doing this you also removed the thread safety of the SDK. I think the issue here is when reconnecting too frequently there are too much active threads. You should investigate on how to handle and clean them.

[alexandrebouthinon]

By removing the thread logic you make the SDK synchrone which is not the purpose. You should maybe handle and clean created thread when reconnecting instead of make it synchrone