feat: parse query comments using custom, faster parser

[levkk]

• Parse query comments using memchr instead of query parser. Use comment-stripped query as the cache key for the query cache, dramatically increasing cache hits on deployments that push request tracing IDs into the queries (e.g. Datadog APM). Not using pg_query::scan also reduces the parsing time considerably.
• Make query fingerprinting lazy. It's not used very much and it's very expensive compared to just query parsing. This reduces overall parsing time by 50%.

[codecov]

Codecov Report

❌ Patch coverage is 97.77283% with 20 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
pgdog/src/frontend/router/parser/comment/query.rs	64.70%	6 Missing ⚠️
pgdog/src/frontend/router/parser/comment/mod.rs	88.37%	5 Missing ⚠️
...og/src/frontend/router/parser/comment/directive.rs	92.30%	4 Missing ⚠️
pgdog/src/frontend/router/parser/cache/ast.rs	98.57%	2 Missing ⚠️
pgdog/src/frontend/router/parser/query/mod.rs	0.00%	2 Missing ⚠️
...dog/src/frontend/router/parser/cache/cache_impl.rs	97.95%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!