Validate fix for transcript multi-word search (#3351)

[Vanashreeshankar]

Draft PR - shared to document investigation; not intended to compete with #3423

While investigating issue #3351, I independently traced the root cause of
the transcript search behavior and implemented a minimal fix locally.
This draft is shared to document findings and the proposed solution.

Problem

Transcript search only worked for single-word queries because matching was done
against individual word spans using the full query string.

Root cause

• In-transcript search operates on individual word spans ([data-word-id])
• Matching the entire query string against a single span causes
  multi-character / multi-word searches (English & Korean) to fail

Solution

• Tokenize query by whitespace
• Match spans that include all query terms
• Preserve existing per-word highlighting and navigation behavior

Result

Multi-character and multi-word searches now work for English and Korean.

Notes

• This PR overlaps with fix: support multi-word transcript search queries #3423, which addresses the same issue
• Opening this as a draft to share investigation details and implementation
• Happy to close or adjust this PR based on maintainers’ preference

Environment note

Full desktop build fails on Windows due to Node memory limits; the change here
is isolated and logic-only.

[netlify]

👷 Deploy request for hyprnote pending review.

Visit the deploys page to approve it

Name	Link
🔨 Latest commit	495c87a

[netlify]

✅ Deploy Preview for hyprnote-storybook canceled.

Name	Link
🔨 Latest commit	495c87a
🔍 Latest deploy log	https://app.netlify.com/projects/hyprnote-storybook/deploys/6989a4688fe20a0008488078