feat(linter): implement noDuplicateProperties

[togami2864]

Summary

closes #2784

Test Plan

Added tests and snapshots

[codspeed-hq]

CodSpeed Performance Report

Merging #4029 will degrade performances by 6.44%

_{Comparing togami2864:2784 (22fd7d6) with main (6e1170e)}

Summary

❌ 2 regressions
✅ 103 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	togami2864:2784	Change
❌	css_analyzer[foundation_11602414662825430680.css]	144.7 ms	154.6 ms	-6.44%
❌	db_17847247775464589309.json[uncached]	15.5 ms	16.5 ms	-6.04%

[dyc3]

Lets try to address the perf regression a little.

[dyc3]

You could add and use an alternative to to_ascii_lowercase_cow in biome_string_case eg to_lowercase_cow

[togami2864]

I think allmost part of this perf regression is caused by the use of the CSS Semantic Model. I'll see if there's room for performance improvement.

[ematipico]

I think allmost part of this perf regression is caused by the use of the CSS Semantic Model. I'll see if there's room for performance improvement.

We should really have a look at that, and understand if the cause is the semantic model or the rule.

You could create a "fake" rule that uses the semantic model, and always trigger the rule.

I believe it's important, because we don't have these kinds of regressions in the JavaScript rules, and when we do, it's usually caused by the business logic of the rule.

[ematipico]

Let's remember the rule pillars: https://biomejs.dev/linter/#rule-pillars

This rule doesn't explain why duplicated properties aren't allowed.

[ematipico]

The text range seems incorrect, did you notice the > character is on line 8? It should not.

Also, It would be nice to add a new note that shows where the first instance of the property is.