fix: keep apostrophes and decode html entities in meta tags

[prakashNalla]

Description

parseAttributes now matches each value up to the same quote it opened with, so content="Don't Panic" and content='The "best" guide' come back in full. decodeEntities is a single pass over named entities (&, <, >, ", ',  ) plus decimal and hex numeric ones, so < becomes < and unknown entities are left as they are. parseMetaTags is exported and the placeholder test is replaced with real cases.

Closes #106

How I Tested This

• npm ci, npm run check (0 errors, 0 warnings, 10 existing hints), npm run test (19 files / 242 tests pass), npm run build, all on Node v22.23.3
• Ran the new parseMetaTags tests against the old parser (with only the export added): 7 of the 8 fail. The unknown-entity case passes on both.
• npm run dev, then /api/preview with the Wikipedia link from the issue returns "ogTitle":"The Hitchhiker's Guide to the Galaxy - Wikipedia" (the live site still returns ').

AI Assistance

Claude helped me understand the issue & I wrote the fix & tests.

Type of Change

• Bug fix
• Tests

Screenshots

N/A, parser change. Covered by the new tests and the /api/preview output above.

Checklist

• This PR targets dev
• I was assigned the linked issue
• npm run check passes
• npm run test passes
• For a bug fix: the new test fails without my change
• Tested locally with npm run dev
• Screenshots included for visual changes
• New templates follow the Template Guidelines

[anurag629]

Thanks @prakashNalla, this covers all of #106, and fast too. I merged it onto current dev: npm run check is clean, all 242 tests pass, and against the old parser with only the export added, 7 of your 8 new tests fail. The unknown entity one passes either way, like you wrote. Merging.