ast_parser: able to define start offset for tokens

[Sysix]

The AST-Parser will parse the source code from the offset 0.
oxc_linter has ParticialLoader which will create multiple source codes for <script> .

At the moment we need to remap the Diagnostic span offset to its original source code offset.
oxc_linter implementation: #9484
oxc_language_server:

oxc/crates/oxc_language_server/src/linter/isolated_lint_handler.rs

Lines 193 to 212 in 1ed18bd

	fn wrap_diagnostics(
	path: &Path,
	source_text: &str,
	reports: Vec<ErrorReport>,
	start: u32,
	) -> Vec<ErrorWithPosition> {
	let source = Arc::new(NamedSource::new(path.to_string_lossy(), source_text.to_owned()));
	reports
	.into_iter()
	.map(|report| {
	ErrorWithPosition::new(
	report.error.with_source_code(Arc::clone(&source)),
	source_text,
	report.fixed_content,
	start as usize,
	)
	})
	.collect()
	}

I would like to give the ast parser an start offset (default 0) which he uses as a start position to generate the Ast::Kind spans.

[Boshen]

I'm against adding a start offset to the parser - it's actually a lot of work.

Let's see whether we can improve the diagnostic reporter instead. Remember we use a fork https://github.com/oxc-project/oxc-miette