refactor(text): parse single-line ATTRIB text via acadrust

Route single-line attribute text through the same acad_text_encode path
TEXT uses (acadrust parse_plain_text, re-encoded for the stroke
tessellator). Multi-line attributes already flow through layout_mtext
(acadrust). No visible change — attribute text with %% codes rendered
correctly before via the local tokenizer; this shares the one parser.
TOLERANCE cells and complex-linetype text still use the local tokenizer
(specialized paths that rarely carry %% codes).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

src/entities/text.rs

@@ -71,7 +71,7 @@ pub struct TextPlacement {
 /// inline grammar: specials arrive resolved to Unicode, and `%%u`/`%%o`
 /// underline/overline become `\L…\l` / `\O…\o` decoration markers. This keeps
 /// TEXT parsing in acadrust rather than OCS's own tokenizer.
-fn acad_text_encode(value: &str) -> String {
+pub(crate) fn acad_text_encode(value: &str) -> String {
     use acadrust::entities::mtext_format::parse_plain_text;
     let doc = parse_plain_text(value);
     let mut out = String::new();