chore: deduplicate GritQL snippets

[arendjr]

Summary

When matching GritQL snippets against a syntax tree, it tries various contexts in which the syntax could occur. These contexts allow it to parse snippets that would not be valid syntax if they were pasted into a top-level source file. Unfortunately, some of them could still end up parsing to the same syntax tree, in which case we end up matching redundant trees. By deduplicating, we can speed up the snippet matching.

Test Plan

Let's see what Codspeed says :)

[codspeed-hq]

CodSpeed Performance Report

Merging #4153 will improve performances by 7.37%

_{Comparing arendjr:deduplicate-snippets (35aa0b4) with main (2f0d5c7)}

Summary

⚡ 1 improvements
✅ 104 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	arendjr:deduplicate-snippets	Change
⚡	db_17847247775464589309.json[cached]	13.8 ms	12.9 ms	+7.37%

[arendjr]

Unfortunately, this PR doesn't show any performance improvements in Codspeed. It could very well be that any improvements are specific to particular queries. This would make sense, since not all queries are parseable in multiple source contexts. In any case, it seems deduplicating is a good thing to do, so I'll merge anyway and maybe it gives some improvements in some situations. Oh well 🤷