fix(starr): Moved webdl and webrip into Amazon regex to prevent incorrect matches

[wheat4714]

Pull Request

Purpose

I created this pull request because I noticed that Amazon CF was applying to certain TV shows and TV episodes. I noticed that the regex is simply looking for "Amazon" in a word boundary. As this is a dictionary word, it's going to match stuff it shouldn't.

Approach

Instead of matching just "Amazon", it requires additional text either side of, which always comes with these releases. Either the resolution or web-rip / web-dl.

https://regex101.com/r/IR8Qbx/6

Requirements

• These changes meet the standards for contributing.
• I have read the code of conduct.

[TRaSH-]

Did a little check what i could find with amazon in the title and added it to your regex test case
https://regex101.com/r/IR8Qbx/7

[wheat4714]

@TRaSH- Better now https://regex101.com/r/IR8Qbx/8

I added a negative lookbehind for "the.amazon" as that seems to be a common pattern, but no sane person would label a streaming service "the streaming service".

I've left all the resolutions in the regex. Removing them would have fixed the incorrect matches as well. Whilst I was not able to find any releases where the resolution is present and WEB-DL/Web-Rip is not, I have only been scrolling through prowlarr results, so cannot be certain that there are none.

[nuxencs]

@wheat4714 this is a way simpler version and it still catches 90% of those cases. the last 10% are just horribly named releases.

https://regex101.com/r/kBRsQW/1

[nuxencs]

please make the same adjustment for the radarr cf

[nuxencs]

please also add back the removed WEBDL and WEBRIP conditions

[bakerboy448]

Reverting this. Breaking change that materially breaks sonarr rename functionality for existing episodes.

This was not actually tested in sonarr and only has blind regex performed