[dajiaohuang]

Summary

• Calculate each parsed node's line number from the source offset where its token begins.
• Add a regression for text containing a newline on the same line as its opening element.

Why

With <root>first line\nsecond line</root>, the text node begins on line 1, but the previous incremental end-offset calculation reported line 2. Using the precomputed line-start offsets preserves the node's actual source line.

Checks

• go test . -run '^(TestLineNumberForMultilineText|TestLineNumbers)$' -count=1
• go test . -v -cover
• gofmt and git diff --check