feat(ingestion): per-district commodity baskets, and a probe that verifies them

Commodities 7 -> 21 and a district_commodity table holding a per-district
basket of 60 pairs, replacing the district x commodity cross product that would
have been 252 mostly-impossible calls a night. District, not zone: Kolar and
Belagavi are both SOUTH and trade almost disjoint baskets.

Only the four original cereals had verified Agmarknet spellings, so the rest
carry sourceVerified = false until POST /admin/ingestion/probe confirms them.
The probe separates a wrong spelling from an absent local market by
cross-referencing which districts and commodities returned rows anywhere — a
distinction that matters because both fail silently and identically.

Three district names were costing us every price in their district: Nasik ->
Nashik, Ahmadnagar -> Ahilyanagar, Belgaum -> Belagavi. The last is the
nastiest: Belgaum still answers with 1000 rows, but they stop in May 2026 while
Belagavi is current, so a retired name serving a frozen archive reads as a
healthy feed. Only the arrival dates give it away.

data.gov.in meters by quota per window, so the limit is reached by the run and
not by any one call: a 62-pair run straight after a 62-pair probe returned 429
on all 62 and recorded the whole ingestion FAILED with zero rows. DataGovClient
now retries with a 45s backoff, long because the window is minutes not seconds.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

src/main/java/com/lincfusion/udyamlinc/master/DistrictCommodity.java

@@ -0,0 +1,64 @@
+package com.lincfusion.udyamlinc.master;
+
+import jakarta.persistence.*;
+import lombok.*;
+
+/**
+ * What a district actually trades — the basket ingestion pulls for it.
+ *
+ * <p>Without this table ingestion is a cross product: every active district
+ * against every active commodity. That is affordable at four cereals and four
+ * districts and stops being affordable the moment fruit and vegetables are
+ * tracked across six states, because most of the pairs it generates cannot
+ * exist. Kolar has no soyabean arrivals, Kamrup no grapes, Nashik no chana.
+ *
+ * <p>The cost is not really the wasted calls, it is what they do to the
+ * freshness screen. Agmarknet answers a filter that matches nothing with
+ * HTTP 200 and an empty record list, so a structurally impossible pair and a
+ * genuinely broken filter produce byte-identical responses. Under a cross
+ * product the majority of pairs return empty on a healthy night, and a
+ * commodity whose {@code sourceCode} is misspelt disappears into that crowd.
+ * Pulling only what a district trades makes an empty result mean something
+ * again: either arrivals genuinely stopped, or something is wrong.
+ *
+ * <p>Zone is deliberately not the grain. Kolar and Belagavi are both SOUTH and
+ * trade almost disjoint baskets — tomato and mango against onion and maize.
+ */
+@Entity
+@Table(name = "district_commodity", indexes = {
+        @Index(name = "ix_dc_district", columnList = "districtId"),
+        @Index(name = "ix_dc_pair", columnList = "districtId,commodityId", unique = true),
+})
+@Getter
+@Setter
+@NoArgsConstructor
+@AllArgsConstructor
+@Builder
+public class DistrictCommodity {
+
+    @Id
+    @GeneratedValue(strategy = GenerationType.IDENTITY)
+    private Long id;
+
+    @Column(nullable = false)
+    private Long districtId;
+
+    @Column(nullable = false)
+    private Long commodityId;
+
+    /**
+     * True where this district is a reference market for the commodity — Nashik
+     * for onion, Kolar for tomato. Not used to decide what to pull; it marks the
+     * rows whose absence is worth a louder complaint, because a national onion
+     * signal that has lost Nashik has lost the market that sets the price.
+     */
+    private Boolean benchmark;
+
+    private Boolean active;
+
+    @PrePersist
+    void onCreate() {
+        if (active == null) active = Boolean.TRUE;
+        if (benchmark == null) benchmark = Boolean.FALSE;
+    }
+}