Update or refresh AWS credentials in an active Pyspark session

So I'm creating and using a SparkSession on Amazon EMR as follows:

os.environ["AWS_ACCESS_KEY_ID"] = access_key_id
os.environ["AWS_SECRET_ACCESS_KEY"] = secret_access_key
os.environ["AWS_SESSION_TOKEN"] = session_token

spark_builder = pyspark.sql.SparkSession \
    .builder \
    .config("spark.hadoop.fs.s3a.aws.credentials.provider",
            "com.amazonaws.auth.DefaultAWSCredentialsProviderChain") \
    .config("spark.driver.extraClassPath",
            "/home/app/lib/hadoop-aws-2.8.4.jar:"
            "/home/app/aws_java_sdk/*") \
    .config("spark.executor.extraClassPath",
            "/home/app/lib/hadoop-aws-2.8.4.jar:"
            "/home/app/aws_java_sdk/*") \
    .config("spark.sql.warehouse.dir", metastore_dir) \
    .config("spark.master", "local[*]") \
    .appName("pyspark explain basic example") \
    .enableHiveSupport()
spark_builder = spark_builder \
    .config("spark.hadoop.fs.s3a.access.key",
            access_key_id) \
    .config("spark.hadoop.fs.s3a.secret.key",
            secret_access_key) \
    .config("spark.hadoop.fs.s3a.session.token",
            session_token) \
    .config("aws.accessKeyId",
            access_key_id) \
    .config("aws.secretAccessKey",
            secret_access_key) \
    .config("aws.sessionToken",
            session_token)
spark = spark_builder.getOrCreate()

del os.environ["AWS_ACCESS_KEY_ID"]
del os.environ["AWS_SECRET_ACCESS_KEY"]
del os.environ["AWS_SESSION_TOKEN"]

This works for, say, accessing a given Amazon s3 location.

I now want to change the AWS credentials within this spark session to access a different s3 location. What's the best way to go about this?

1. Looking at SparkSession pages, I don't see an obvious way to update the runtime configuration. There's newSession method but that doesn't accept any conf parameter (or any parameter, for that matter).
2. While writing this, I came across this SO thread which suggests stopping and starting spark session since SparkSession and SparkContext are not modifiable after start.

New stackoverflow poster, please ignore any etiquette misses :)