fix: handle malformed og:url protocols missing colon

[rmdes]

Some websites return og:url values like 'https//example.com' (missing
colon after protocol). These get treated as relative URLs and
concatenated with the base URL, creating broken links.

This fix:

• Adds fixMalformedUrl() to correct 'https//' -> 'https://'
• Improves URL validation regex to require proper protocol
• Validates the corrected URL before using it

Problem

Some websites have malformed og:url meta tags like https//example.com (missing the colon after the protocol). When the Open Graph scraper encounters this, it treats it as a relative URL and resolves it against the base URL, creating broken links like:

https://www.lecho.behttps//www.lecho.be/path

Example of affected post: https://bsky.app/profile/polbegov.skyfleet.blue/post/3maw2bc3deb25

Root Cause

The malformed og:url (e.g., https//www.lecho.be/path) doesn't match the protocol pattern, so URL resolution treats it as a relative path and prepends the base domain.

Solution

1. Added fixMalformedUrl() function that corrects common protocol typos:

   • https// → https://
   • http// → http://

2. Improved the URL validation regex:

   • Now requires a proper protocol (was previously optional)
   • Validates the corrected URL, falling back to the original RSS URL if still invalid