fix: use character count for regex lint spans (fixes #100)

[Kubudak90]

Problem
The generic Regex lint in eipw-lint/src/lints/markdown/regex.rs uses byte offsets directly when creating annotation spans. When the source text contains multi-byte UTF-8 characters (e.g., emojis, non-ASCII text), the byte offset exceeds the character count expected by the annotation framework, causing a panic:

SourceAnnotation range is bigger than source length

This is the same root cause as #100, but affects all markdown lints that use the Regex struct (e.g., markdown-link-first, markdown-link-status, markdown-rel-links, etc.), not just markdown-json-cite.

Fix
Convert byte offsets to character counts using source[..start].chars().count() before creating annotation spans, matching how the linting framework expects positions.

Files changed

• eipw-lint/src/lints/markdown/regex.rs: 6 insertions, 2 deletions

Verification

• cargo test passes
• The fix follows the same pattern already used in json_schema.rs (PRs fix: prevent panic in markdown-json-cite with non-ASCII input #162/fix: use char count instead of byte length for json_schema spans #164)

Fixes #100