psqldef: Fix handling duplicated PostgreSQL OID from extension

[chumaltd]

Hi, I just implemented more accurate extension detection logic.

PostgreSQL's oid is not globally unique, and you need to search both with its own oid and oid of its object class.
I hit into a conflict case between my table oid & extension's function oid.

Past releases of psqldef after e57a042 can eventually see this issue after long run with extensions.

I just confirmed go test finished successfully under cmd/psqldef.
I'm sure this patch fits original intension of e57a042, but I have no ideas about:

• pg_type may require handling pg_enum, too. But WHERE relname IN ('pg_type', 'pg_enum') will be still ambiguous between their combinations. I took just mitigation at this point.
• We don't have corresponding test cases simulating oid confliction.

I can test with my production DBs when #1060 quick solution will be landed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
database/postgres/database.go	81.92% <100.00%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[gfx]

Thanks!

Would be nice if we have at least one test case for this, but for now, I'm merging this PR and will immediately release a new version of sqldef.