feat(NODE-7097): adopt built-in zstd support

[axelonet]

Description

Summary of Changes

• Prefer Node.js built-in zstd support when available
• Fall back to @mongodb-js/zstd when built-in support is unavailable
• Update zstd compression tests to cover driver round-tripping and optional dependency behavior

Release Highlight

Release notes highlight

Double check the following

• Lint is passing (npm run check:lint)
• Self-review completed using the steps outlined here
• PR title follows the correct format: type(NODE-xxxx)[!]: description
  • Example: feat(NODE-1234)!: rewriting everything in coffeescript
• Changes are covered by tests
• New TODOs have a related JIRA ticket

[PavelSafronov]

@axelonet , thanks so much for your contribution. The team will triage this PR and https://jira.mongodb.org/browse/NODE-7097 and will get back to you.

[sarthaksoni25]

Hi, I'm new to this codebase and have been reading through some open PRs to learn, this one caught my attention. Please correct me if I'm misreading anything.

• On the test in optional_require.test.ts - the new test covers the case where the addon is absent and built-in is used instead. But since CI explicitly installs @mongodb-js/zstd via run-tests.sh, this test would always skip there. The priority behaviour i.e. built-in preferred over addon when both are present, doesn't seem to have test coverage. Is that intentional?

• On the compression test in compression.test.ts - I'm new to zstd so correct me if I'm wrong. Reading through the code, it looks like on Node 22.15.0+ both compress() and decompress() end up using the same built-in implementation. Doesn't that make the round-trip test self-referential? I see zstd frames always start with magic bytes 0xFD2FB528 per the spec, would checking those bytes be a simpler independent verification?

Happy to be corrected.