fix(rest/python): rebuild discounts.applied idempotently on recalculation

[XiaolongZhang-TT]

Description

_recalculate_totals rebuilt checkout.totals from scratch on every create/update but appended to discounts.applied without resetting it. Because the persisted checkout (reloaded on each update) already carries the applied entries from the previous response, an update that omits the discounts field accumulates a duplicate applied entry on every recalculation.

Repro: create a checkout with discounts.codes = ["10OFF"] (response applied has 1 entry), then PUT /checkout-sessions/{id} with a body that omits discounts (e.g. a quantity / shipping change) → the response discounts.applied lists the discount twice, and keeps growing with each such update.

Root cause — rest/python/server/services/checkout_service.py, _recalculate_totals: checkout.totals is reset (checkout.totals = []) at the top of each recalculation, but discounts.applied is never reset, so it accumulates once the reloaded checkout brings back the prior entries.

Fix: reset checkout.discounts.applied = None before recomputing — the same "server is the authority, rebuild from scratch" pattern already used for totals. This matches discount.json, where applied carries ucp_request: "omit" (a server-authoritative output field), so it must be rebuilt idempotently rather than appended to.

Category (Required)

• Samples / Conformance: Maintaining samples and the conformance suite. (Requires Maintainer approval)

Related Issues

None — this is a newly identified bug in the Python sample server (unrelated to #121 / #134).

Checklist

• I have followed the Contributing Guide (including Conventional Commits title requirements and ! for breaking changes).
• I have updated the documentation (if applicable). — N/A, no documentation changes.
• My changes pass all local linting and formatting checks. (ruff check + ruff format clean)
• I have added tests that prove my fix is effective or that my feature works. (test_discount_applied_is_not_duplicated_on_update)
• New and existing unit tests pass locally with my changes. (Python integration suite: 12 passed)
• (For Core/Capability) I have included/updated the relevant JSON schemas. — N/A, no schema changes.
• I have regenerated Python Pydantic models by running generate_models.sh under python_sdk. — N/A, no schema changes.

Screenshots / Logs (if applicable)

Before/after from the regression test — discounts.applied after an update that omits the discounts field:

Before (bug):

applied == ["10OFF", "10OFF"]   # duplicated on each update

After (fix):

applied == ["10OFF"]            # rebuilt idempotently

Full Python integration suite:

12 passed, 2 warnings in 1.76s

[XiaolongZhang-TT]

Marking this ready for review. Quick context for reviewers:

• Minimal bugfix in the Python sample server (Samples / Conformance).
• Root cause: _recalculate_totals rebuilt totals from scratch each create/update but appended to discounts.applied without resetting it, so an update that omits the discounts field duplicated the applied entries (they come back from the persisted checkout).
• Fix: reset discounts.applied before recomputing — same "server is authority, rebuild from scratch" pattern already used for totals. Matches discount.json (applied is ucp_request:"omit").
• Regression test added; it fails before the fix (applied == ["10OFF","10OFF"]) and passes after. Full Python integration suite passes locally (12/12); ruff clean.
• No schema / protocol / route changes. Unrelated to Flower Shop verify_signature accepts every request and its docstring describes HMAC — spec requires RFC 9421 asymmetric signatures #121 (signatures), [Feat]: implement the Cart capability in the Python sample server #134 (cart). fix(rest/python): seed valid totals when creating a checkout #152 touches the same file but a different section (create-path totals seeding) — happy to rebase if that lands first.

Open to a different reset location if a reviewer prefers it.