feat: add PostgreSQL archive backend

[yunwei37]

Closes #723.

Summary

• add an opt-in PostgreSQL backend through -database-url or SLACKDUMP_DATABASE_URL
• keep SQLite as the default with no behavior change when the option is unset
• add PostgreSQL migrations and the dialect-specific SQL needed by archive/resume
• redact database URLs from stored command arguments
• keep downloaded files on the filesystem
• reject -dedupe on PostgreSQL because the existing dedupe implementation is SQLite-specific
• exercise archive and resume against PostgreSQL 17.6 in CI

Validation

• go test ./...
• go test -race -cover ./...
• SLACKDUMP_POSTGRES_TEST_URL=... go test ./internal/chunk/backend/dbase -run TestPostgresArchiveResume -v
• built and exercised archive/resume with a real PostgreSQL database in a containerized deployment

Scope

This PR intentionally covers the archive/resume storage path only. SQLite remains supported and remains the default. PostgreSQL credentials are supplied at runtime and are not emitted in help defaults or persisted in session arguments.

[rusq]

Hey @yunwei37, thanks a lot for the contribution. I'll review this when able.

[rusq]

Thanks again for taking on PostgreSQL support — the implementation is useful and provides a good reference for the database-specific pieces.

I think PostgreSQL needs to be integrated as a first-class archive backend across the command surface: view, export, dump, archive, resume, and applicable tools commands such as dedupe, rather than introduced through command-specific branches in only archive and resume.

That likely means extending the existing source/archive abstractions so commands operate on a backend-neutral target, with SQLite and PostgreSQL providing their respective implementations. I may rework this in that direction, using this PR as a reference where helpful.