CalicoParser now tracks how many rows it drops for a blank/N-A covering type
and how many references it drops for not resolving to a portfolio UPRN, exposed
for the load reconciliation report so nothing vanishes silently.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The real Calico export lodges the literal string "N/A" (not an empty cell) for
placeholder rows; _clean_str now treats N/A-family sentinels as no covering, so
the 1,239 placeholders are dropped on parse (5,346 -> 4,107 real observations).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
CalicoParser reads the workbook into CalicoAssetCondition rows, drops rows
with no covering type, and groups them by asset reference into CalicoProperty,
resolving each reference to a UPRN via the injected lookup (unmatched dropped).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Behaviour-preserving lift-and-shift of the condition module out of legacy
backend/condition/ into domain/condition, infrastructure/condition,
infrastructure/postgres/condition_tables.py, repositories/condition,
applications/condition, and tests/condition. Imports rewritten to the DDD
paths; ConditionPostgres and the ORM models keep the legacy backend.app.db
Base/db_session bridges so the existing suite proves behaviour is unchanged
(SQLModel + session-DI conversion tracked as a follow-up in ADR-0064).
16 condition tests pass at the new location.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
