lexical-ios

Document all functions within ElementNode. (You can copy descriptions from the Lexical JS docs for anything where it makes sense to do so.) Then make a markdown file to categorise the methods into topics. (See Node.md for an example how to do this. It uses Apple's DocC format.)

Two fixes that allow editor states containing code blocks to round-trip through EditorState.fromJSON: CodeNode and LineBreakNode are built-in node types with public NodeTypes and full rendering support, but neither was present in Editor's default registeredNodes map. As a result, deserializing any editor state that contains a code block (or a bare line break) could not reconstruct those nodes. Register them alongside the other built-ins (root, text, element, heading, paragraph, quote). CodeNode's Decodable init required a "language" key. Editor states produced by other Lexical implementations (e.g. lexical-web) omit "language" when no language is set, so decoding threw keyNotFound. Use decodeIfPresent with an empty-string default so a missing language is tolerated.

Instead of appending a separate indent attribute for each block-level ancestor, accumulate all indent values into a single totalIndent and emit one .indent_internal attribute. This prevents child node indent values (e.g. ParagraphNode's 0) from overwriting parent indent values (e.g. QuoteNode's 1) during attribute merging. Adds regression test testParagraphInsideQuoteRetainsIndent verifying that text inside a QuoteNode receives the correct headIndent and firstLineHeadIndent from the quote's indent level. This brings the quote node behavior of lexical-ios to match the behavior of lexical web.