Avoid splitting query with filtered include Take(1)

Start by reproducing the query using AsSplitQuery, Include, Take(1), and ToArray with the supplied PostgreSQL model and SQL. Trace the EF Core query translation and split-query entry points to understand why the filtered include is split. Done means the filtered include can use a single join where appropriate while split queries still avoid cartesian explosion for additional includes, with coverage for the generated SQL.

public class Post
{
    public int Id { get; set; }
    public string Name { get; set; }
    public List<Author> Authors { get; set; }
}

public class Author
{
    public int Id { get; set; }
    public string Name { get; set; }
}

public class MyDbContext : DbContext
{
    public static readonly ILoggerFactory LoggerFactory = LoggerFactory.Create(b => b.AddConsole());

    public DbSet<Post> Posts { get; set; }
    public DbSet<Author> Authors { get; set; }

    protected override void OnConfiguring(DbContextOptionsBuilder options) =>
        options.UseNpgsql("User ID=postgres;Password=root;Host=localhost;Port=5432;Database=test")
            .UseLoggerFactory(LoggerFactory);
}

await using (var db = new MyDbContext());
    Post[] posts = db.Posts
        .AsSplitQuery()
        .Include(p => p.Authors.Take(1))
        .ToArray();

The above code using AsSplitQuery generates the two queries:

SELECT p."Id", p."Name" FROM "Posts" AS p ORDER BY p."Id";

SELECT t0."Id", t0."Name", t0."PostId", p."Id"
FROM "Posts" AS p
INNER JOIN (
    SELECT t."Id", t."Name", t."PostId"
    FROM (
        SELECT a."Id", a."Name", a."PostId", ROW_NUMBER() OVER(PARTITION BY a."PostId" ORDER BY a."Id") AS row
        FROM "Authors" AS a
    ) AS t
WHERE t.row <= 1) AS t0 ON p."Id" = t0."PostId"
ORDER BY p."Id", t0."PostId", t0."Id";

when only one with a join could be made because of the Take(1), something like:

SELECT p."Id", p."Name", a."Id", a."Name"
FROM "Posts" AS p 
JOIN "Authors" AS a ON p."Id" = a."PostId"
ORDER BY p."Id", a."Id";

One could say don't use AsSplitQuery but let's assume I'm also including all comments of a post, splitting query would help avoid cartesian explosion.

Also the generated join seems very complex, is it expected or should I open another issue?