fix(spp_change_request_v2): let the registrant prefill through the freeze on an empty repaired detail (#443)

[gonzalesedwin1123]

Part of #443 (weekly ci-full red). Second per-module PR; a real production bug, not a test problem.

Problem

19.0.3.1.10 let _ensure_detail() bind a new detail row to a submitted change request (a request that never got one, or lost it, could not be opened from any context). The binding now passes, but _ensure_detail() then calls detail.prefill_from_registrant(), which writes the registrant's current values to the mapped fields, and _assert_content_editable refuses that write whenever the request type has field_mapping mappings. A fresh detail holds False everywhere, the prefill values are truthy, so "differs" → UserError: This change request has already been submitted for approval, so its proposed changes are locked.

Every shipped Edit Individual / Edit Group type carries those mappings through spp_cr_types_base, which every real deployment installs. The module's own test type is built from CR_TYPE_DEFS with no mappings, so the protected set was just {field_to_modify} and TestFrozenDetailBinding passed in per-module CI while failing in the weekly SP-MIS stack. action_open_detail / action_goto_details both go through _ensure_detail(), so on a real deployment the repair 19.0.3.1.10 set out to do still does not work.

Fix

The prefill moves into the detail's create(): _ensure_detail builds the row with _prefill_values() (the registrant's current values for the prefill mapping, record values as ids) already in the create vals. The repair therefore performs no post-submit write at all, and _assert_content_editable keeps no exemption: every write to a mapped field on a submitted request, the registrant's own value included, stays refused.

Why the row must be prefilled rather than left empty: field_mapping.apply writes empties on purpose ("a user may be intentionally clearing a field"), so approving an empty repaired detail would clear every mapped field on the registrant. Prefilled, it equals what the registrant already holds and apply skips it.

The first draft of this PR opened a shape-based carve-out in the freeze instead ("empty detail + registrant's own values"). The adversarial review showed it validated the no-op claim against the prefill mapping while apply uses the apply mapping (target_field, transform), and that its precondition is exactly the shape of an approved clear-every-field proposal, so it left a one-write slot. The create-time prefill removes that surface entirely.

prefill_from_registrant() keeps its public behaviour (a write, therefore subject to the freeze on submitted requests). spp_change_request_v2 19.0.3.1.16 → 19.0.3.1.17 + HISTORY. README/index.html left for CI's generator.

Tests (tests/test_frozen_detail_binding.py, 6 → 11)

The class now gives the test-created edit_individual / edit_group types the mappings spp_cr_types_base ships, so module CI exercises the real shape. Against the unfixed model that alone turns the repair tests red in module-only mode. New:

• Edit Group can be repaired after submit (fully overlapping prefill/apply mappings too);
• the repaired detail arrives already populated with the registrant's values, without any write;
• prefill_from_registrant() on an empty submitted detail is refused (the freeze keeps no exemption);
• another value on an empty submitted detail is refused;
• restoring a single cleared field to the registrant's own value is refused.

Evidence: with -i spp_change_request_v2,spp_cr_types_base --test-tags /spp_change_request_v2:TestFrozenDetailBinding the class went from 0 failed, 2 error(s) of 6 to 0 failed, 0 error(s) of 11; module suite 0 failed, 0 error(s) of 425 tests. All pre-commit hooks pass. An SP-MIS ci-full is dispatched on this branch; its failing set should lose exactly the two TestFrozenDetailBinding errors against the 2026-09-20 baseline.

Pre-existing gaps found in review (not this PR's scope; filed separately)

spp.change.request.registrant_id is not among the fields frozen on submit; create() on detail models is unguarded, so a populated row can be crafted and bound to a detail-less submitted request through the 19.0.3.1.10 binding rule; Studio-generated detail models inherit neither the freeze nor the prefill.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.91%. Comparing base (1a3c591) to head (d90820c).

Additional details and impacted files

@@           Coverage Diff           @@
##             19.0     #536   +/-   ##
=======================================
  Coverage   76.91%   76.91%           
=======================================
  Files         704      704           
  Lines       45774    45784   +10     
=======================================
+ Hits        35205    35216   +11     
+ Misses      10569    10568    -1     

Flag	Coverage Δ
spp_api_v2_change_request	73.37% <ø> (ø)
spp_base_common	91.07% <ø> (ø)
spp_change_request_v2	79.01% <100.00%> (+0.11%)	⬆️
spp_cr_type_assign_program	92.50% <ø> (ø)
spp_dci_demo	94.28% <ø> (ø)
spp_farmer_registry_cr	61.24% <ø> (ø)
spp_farmer_registry_demo	63.39% <ø> (ø)
spp_mis_demo_v2	70.41% <ø> (ø)
spp_programs	67.58% <ø> (ø)
spp_registry	89.00% <ø> (ø)
spp_security	69.56% <ø> (ø)
spp_starter_sp_mis	86.84% <ø> (-2.05%)	⬇️
spp_studio_change_requests	84.85% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
spp_change_request_v2/models/change_request.py	84.70% <100.00%> (+0.27%)	⬆️
...ge_request_v2/models/change_request_detail_base.py	75.00% <100.00%> (+0.54%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gonzalesedwin1123]

Full-stack evidence: SP-MIS ci-full dispatched on this branch (run 35565791203, at 440596d) vs the 2026-09-20 weekly baseline (run 35499069639).

• Result line: 11 failed, 9 error(s) of 7405 vs baseline 11 failed, 11 error(s) of 7397 (the 8 extra tests are this PR's).
• Failing-test set diff: gone TestFrozenDetailBinding.test_ensure_detail_can_bind_after_submit, TestFrozenDetailBinding.test_get_detail_works_after_repair; new: none. The remaining failures are the other ci-full: 23 tests red under the full demo stacks (2 production bugs + 9 test-isolation causes); originally the spp_registry unlink tests since #411 #443 causes, each with its own PR.