Three difference approaches to configure Spring JPA + Hibernate 5 - what's the difference?

Question

I'm upgrading an existing project to use Spring Data JPA and repositories instead of a few DAOs which query data using Hibernate.

So I managed to create a valid configuration which works:

public class PersistenceConfiguration
{
    (...)

    public DataSource dataSource()
    {
        ComboPooledDataSource dataSource = new ComboPooledDataSource();

        (...)

        return dataSource;
    }

    @Bean( "jpaProperties" )
    public Properties jpaProperties()
    {
        Properties properties = new Properties();

        (...)

        return properties;
    }

    @Bean(name = "entityManagerFactory")
    @Primary
    public LocalContainerEntityManagerFactoryBean entityManagerFactory()
    {
        LocalContainerEntityManagerFactoryBean factory = new LocalContainerEntityManagerFactoryBean();
        factory.setPersistenceUnitName( "entityManagerFactory" );
        factory.setPackagesToScan( "my.package.domain" );
        factory.setJpaVendorAdapter( getHibernateJpaVendorAdapter() );
        factory.setJpaProperties( jpaProperties() );
        factory.setDataSource( dataSource() );

        return factory;
    }

    @Bean( name = "transactionManager" )
    public PlatformTransactionManager transactionManager( EntityManagerFactory entityManagerFactory )
    {
        JpaTransactionManager txManager = new JpaTransactionManager();
        txManager.setEntityManagerFactory( entityManagerFactory );
        return txManager;
    }

    protected HibernateJpaVendorAdapter getHibernateJpaVendorAdapter()
    {
        HibernateJpaVendorAdapter vendorAdapter = new HibernateJpaVendorAdapter();
        vendorAdapter.setGenerateDdl( false );
        vendorAdapter.setDatabase( (this.databasePlatform.contains( "MySQL" ) ? Database.MYSQL : Database.H2) );
        vendorAdapter.setDatabasePlatform( this.databasePlatform );
        return vendorAdapter;
    }

    @Bean
    public PersistenceExceptionTranslationPostProcessor exceptionTranslation()
    {
        return new PersistenceExceptionTranslationPostProcessor();
    }
}

As you can see I have used LocalContainerEntityManagerFactoryBean - I have followed a Baeldung tutorial and LocalEntityManagerFactoryBean is another option but tutorial's author claims it is "less flexible". Then I have started digging deeper just to understand how Spring Data JPA works and I found out that there is a third way - LocalSessionFactoryBean.

I guess I could not bother and just stick to whatever works, but I'm quite new to Spring and I would like to avoid copy-pasting code without at least trying to understand it.

Here is what I have found out so far:

LocalSessionFactoryBean:

If I use it, I can access SessionFactory and query data using Hibernate.

Official documentation says:

... and Hibernate BeanContainer integration will be registered out of the box.

Is it what make SessionFactory bean available? I shouldn't use LocalSessionFactoryBean if I don't need accessing SesionFactory?

LocalContainerEntityManagerFactoryBean vs LocalEntityManagerFactoryBean:

I have found a discussion with the explaination how they differ here, but even though a few concepts there aren't clear to me, I understand that for my small self-contained I could stick to LocalEntityManagerFactoryBean. Quick change from to LocalEntityManagerFactoryBean reveals that it doesn't have #setPackagesToScan and #setDataSource methods.

For former, does it mean that the whole app's package tree is being scanned automatically?

What's about #setDataSource - will it just pick up DataSource bean automatically or should I just stick LocalContainerEntityManagerFactoryBean?

What's the purpose of LocalEntityManagerFactoryBean in the first place?