Java SDK connection pool exhaustion despite explicit configuration

Looking for advice on why the Genesys Cloud Java SDK is not respecting our custom HTTP client configuration for connection pooling. We are building a high-throughput token service and need to manage thread-safe HTTP connections efficiently. According to the documentation, we can pass a pre-configured ApacheHttpClient to the ApiClient constructor.

Here is our initialization code:

PoolingHttpClientConnectionManager connectionManager = new PoolingHttpClientConnectionManager();
connectionManager.setMaxTotal(200);
connectionManager.setDefaultMaxPerRoute(50);

CloseableHttpClient httpClient = HttpClients.custom()
 .setConnectionManager(connectionManager)
 .build();

ApiClient client = ApiClient.newBuilder().setHttpClient(httpClient).build();

However, under load, we see ConnectionPoolTimeoutException errors when calling /api/v2/oauth/token. The SDK seems to be ignoring the pool settings or creating new clients per request internally.

We are sending this payload:

{
 "grant_type": "client_credentials",
 "client_id": "xxx",
 "client_secret": "yyy"
}

Why does the SDK not adhere to the provided HttpClient instance? Is there a specific method to ensure thread-safety across multiple worker threads using the same ApiClient instance?