Remove duplicate file preservation after import workflow matures

[codeforester]

Summary

Once duplicate-import handling is proven in regular use, revisit the temporary duplicates/ preservation policy introduced by #43. The mature behavior should likely skip exact duplicate successful imports and remove the inbox copy after recording history, without keeping another physical copy.

Background

Issue #43 intentionally preserves exact duplicate inbox files in a managed duplicates/ directory. That is conservative while BankBuddy is young and the import/archive workflow is still earning trust. Long term, exact SHA-256 duplicates of already successful imports do not need duplicate storage because the primary canonical processed copy already exists.

Future desired behavior

• Keep SHA-256-based exact duplicate detection.
• Keep auditable duplicate/skipped import history.
• Stop preserving physical duplicate files once the workflow is mature enough.
• Remove or deprecate the managed duplicates/ directory policy.
• Update docs, spec, and tests to reflect the simplified lifecycle.

Acceptance criteria

• Exact successful duplicate inbox files are removed after duplicate history is recorded.
• The existing canonical processed file remains the only retained physical statement copy.
• import history still explains the skip clearly.
• Any duplicate-path schema or output fields introduced for Skip exact duplicate inbox imports and preserve duplicate files #43 are removed, deprecated, or clearly documented as historical-only.
• README and design spec no longer recommend preserving duplicate files.

Depends on

• Skip exact duplicate inbox imports and preserve duplicate files #43