Refactor Lexbor empty text detection to ASCII whitespace only, changing it from Lexbor's original lxb_dom_node_is_empty check!

[pygarap]

This pull request updates the definition of "empty" text nodes in the Lexbor backend of selectolax, changing it from Lexbor's original lxb_dom_node_is_empty check to a stricter check for nodes containing only ASCII whitespace (space, tab, newline, form feed, or carriage return). It also adds a new test to verify this behavior on a full HTML document and updates documentation throughout the codebase to reflect the new definition.

Important

lxb_dom_node_is_empty is overkill for our needs, and it also causes some bugs with the previous implementation.

Behavioral changes:

• The _is_empty_text_node method in LexborNode now checks if a text node consists solely of ASCII whitespace, rather than using Lexbor's lxb_dom_node_is_empty. This is implemented via a new helper method _is_whitespace_only, which does an inline scan for whitespace characters. [1] [2]
• All code and documentation references to skipping or identifying empty text nodes now specify that only ASCII whitespace nodes are considered empty, rather than relying on Lexbor's broader or different definition. [1] [2] [3] [4] [5] [6] [7] [8] [9]

Testing:

• A new test, test_traverse_with_skip_empty_on_a_full_html_document, is added to verify that traversal with and without skip_empty=True produces the expected results, confirming the new whitespace-only logic.
• Minor utility code is added to clean up docstrings for test inputs.

These changes make the definition of "empty" text nodes more predictable and easier to reason about, matching common expectations for HTML whitespace handling.