markdown: Fix markdown copy range resolution

[pkondzior]

Summary

Markdown copy could break formatting when selection edges landed inside wrapper syntax, producing output like bold** instead of **bold**. This PR makes copy range resolution span-aware, so copied markdown keeps valid boundaries.

Why

The old flow used raw source slicing. That works for plain text but is fragile for formatted markdown. Using parse-time span metadata makes selection-to-source mapping consistent for wrappers like bold, links, and inline code.

What changed

The markdown parser now keeps wrapper bounds for inline code, and markdown copy resolution now runs through source-span normalization and wrapper expansion logic. Copy still falls back to clamped raw selection when parsed source is stale. Tests were expanded across parser, resolver, and markdown integration paths.

Cases covered

Case	Description	Before	After
1. Full formatted token selection	Selecting visible text inside a full formatted token now copies the full markdown wrapper.
2. Partial formatted token selection	Selecting only part of formatted content now copies only that part, without leaking wrapper syntax.
3a. Link boundary handling	Selections near link edges no longer leak broken suffixes like ](...).
3b. Link boundary handling (additional)	Additional boundary scenario for links now resolves cleanly.
4. Multiline list boundary handling	Selections spanning list items now start cleanly and do not inherit suffix fragments from prior items.
5. List boundary with inline style wrapper	If a selection begins at a list-item boundary inside visible styled text, copy now preserves complete inline style wrappers.

Closes #42958

• Added solid test coverage and/or manual testing screenshots
• Completed self-review, including security and performance considerations
• Aligned UI changes with the UI/UX checklist

Release Notes:

• Fixed markdown copy behavior for formatted selections.