fix: serialize None as null keyword in PDFArray

[agustin18]

Description

Per ISO 32000-1 § 7.3.9 ("Null Object"):

The null object has a type and value that are almost never equal to any other object; there is only one object of this type, denoted by the keyword null.

While render_pdf_primitive in fpdf/syntax.py correctly converts None to "null", PDFArray.serialize() previously omitted None handling and fell back to Python's str(elem), producing invalid PDF tokens like [None] or [None None].

This PR adds explicit serialization of None as "null" in PDFArray.serialize(), handling both homogeneous arrays ([None, None] -> [null null]) and heterogeneous arrays ([1, None] -> [1\nnull]).

Tests

• Added parameterized unit tests in test/test_syntax.py covering homogeneous null arrays, single-item null arrays, and heterogeneous arrays mixed with integers, booleans, and indirect objects (PDFObject).