fix(profiling): prevent integer overflow in Invalid_Zip3_USA prevalence formula

The dq_score_prevalence_formula cast record_ct (BIGINT) to INT, overflowing
on tables with more than 2.1B rows and aborting hygiene issue detection for
the whole profiling run. Drop the cast so the numerator stays BIGINT, matching
the other operand and the Column_Pattern_Mismatch formula. Reaches existing
deployments via the static-metadata refresh on upgrade; no migration needed.

TG-1120

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: aarthy-dk

testgen/template/dbsetup_anomaly_types/profile_anomaly_types_Invalid_Zip3_USA.yaml

@@ -16,7 +16,7 @@ profile_anomaly_types:
   suggested_action: |-
     Review your source data, ingestion process, and any processing steps that update this column.
   dq_score_prevalence_formula: |-
-    (NULLIF(p.record_ct, 0)::INT - NULLIF(SPLIT_PART(p.top_patterns, ' | ', 1), '')::BIGINT)::FLOAT/NULLIF(p.record_ct, 0)::FLOAT
+    (NULLIF(p.record_ct, 0) - NULLIF(SPLIT_PART(p.top_patterns, ' | ', 1), '')::BIGINT)::FLOAT/NULLIF(p.record_ct, 0)::FLOAT
   dq_score_risk_factor: '1'
   dq_dimension: Validity
   impact_dimension: Conformance