Fix NVDA+K not recognizing links with nested elements (#19428)
Fixes #17363
### Summary of the issue:
Previously, the `NVDA+K` command failed to identify links when the caret
was positioned on text wrapped in nested elements (e.g., `<strong>`,
`<em>`, or `<span>`) within an `<a>` tag. This occurred because the
command only checked the immediate object at the caret or its direct
parent, missing links further up the hierarchy.
### Description of user facing changes:
`NVDA+K` now correctly identifies and reports the destination of links
even when the text at the caret position is formatted (e.g., bold or
italicized) or contained within other nested elements.
### Description of developer facing changes:
In `source/textInfos/__init__.py`:
- Updated `_getLinkDataAtCaretPosition` to traverse up the object
hierarchy (max 10 levels) to find an enclosing link if the current
object is not one.
- Added a docstring to `_getLinkDataAtCaretPosition`.
- Refactored the logic to explicitly handle graphics inside links
(preserving existing fix #14779) before checking for general link
ancestors.
### Description of development approach:
The solution involves checking the ancestors of the object at the caret
position. If the current object is not a link, the code iterates up the
parent chain until a link is found or the traversal limit (10 levels) is
reached. This ensures robust detection of enclosing links while
preventing infinite loops in case of malformed object trees.
### Testing strategy:
- Verified `NVDA+K` on an HTML test case with links containing nested
`<strong>`, `<em>`, and `<span>` tags.
- Verified that `NVDA+K` still works for simple links.
- Verified regression testing for graphics inside links (as per #14779).
### Known issues with pull request:
None.
