Implement Regional Sequence Generators for User Table

[Suhasrv2403]

Summary

This PR implements the multi-region sequence generator changes required to support GDPR-compliant user data residency across Virginia, Ireland, and Stockholm. It introduces a new user_id format, an is_eu flag, and disaster recovery sequence ranges for all three regions.

---

Problem

The existing generate_sid() function had the following issues:

• Used INT instead of BIGINT, limiting scale
• Old 3-segment SID-00-XXX-XXX-XXX format had no region identity
• No is_eu awareness — EU and non-EU users were treated identically
• No disaster recovery sequence ranges — ID collisions would occur during regional failover
• Sequence was unqualified and unbound (NO MINVALUE, NO MAXVALUE)

---

Approach

New ID format:
SID-001-XXX-XXX-XXX-XXX → non-EU users
SID-002-XXX-XXX-XXX-XXX → EU users

Sequence range partitioning:

Region	Role	Prefix	Range	Direction
Virginia	Normal non-EU	SID-001	1 → 1T	Ascending
Ireland	DR Virginia	SID-001	1T → 1	Descending
Ireland	Normal EU	SID-002	2T → 4T	Ascending
Stockholm	DR Ireland	SID-002	4T → 2T	Descending
Virginia	DR EU both	SID-002	5T → 4T	Descending

Why directional sequences for DR?

• Normal operation counts ascending from the bottom of its range
• DR operation counts descending from the top of its range
• The two can never collide within a realistic DR window
• On recovery and re-sync, origin of any ID is traceable by whether it is high or low within its range

is_eu flag:

• Added to all three regions' users tables
• Virginia defaults FALSE, Ireland and Stockholm default TRUE
• Virginia and Stockholm are each dedicated to one role — no branching needed
• Ireland branches on is_eu to pick between its two sequences

---

Files Changed

File	Purpose
migrate_virginia_users.sql	One-time migration for existing Virginia rows
migrate_ireland_users.sql	One-time migration for existing Ireland rows
migrate_stockholm_users.sql	One-time migration for existing Stockholm rows
ddl_virginia_users.sql	Clean schema definition for Virginia
ddl_ireland_users.sql	Clean schema definition for Ireland
ddl_stockholm_users.sql	Clean schema definition for Stockholm
test_virginia_generate_sid.sql	pgTAP tests for Virginia
test_ireland_generate_sid.sql	pgTAP tests for Ireland
test_stockholm_generate_sid.sql	pgTAP tests for Stockholm

---

Migration Order

1. Run migration scripts first — these create sequences and backfill existing rows
2. Run DDL scripts second — these define the clean schema for new deployments
3. Run tests third to verify all regions generate correct IDs

Closes #238

[Suhasrv2403]

Updates Based on Review Comments

Changes Made

1. Ireland EU prefix corrected
Fixed Ireland EU users to use SID-EU- prefix instead of SID-002-. This was missed in the original submission.

2. Updated sequence ranges
Moved from trillion-scale ranges to billion-scale ranges as discussed:

Region	Role	Prefix	Range
Virginia	Normal non-EU	SID-001	1 → 20B
Ireland	DR Virginia	SID-001	20B → 40B
Stockholm	DR Virginia	SID-001	40B → 60B
Ireland	Normal EU	SID-EU	1 → 20B
Stockholm	DR Ireland	SID-EU	20B → 40B

3. Math extract approach adopted
Region identity is derived from the sequence value itself — dividing by 20B naturally produces the region segment (000, 001, 002). Same function works identically across all regions with no hardcoding needed.

4. Stockholm DDL added
Stockholm was missing from the EU DDL file. Added at the end of ddl_eu_users.sql using Ireland as reference, including table definition, sequences, generate_sid() function and trigger.

5. Migration scripts removed
Removed as discussed — scope is DDL and function changes only.

---

Files Changed

File	Change
ddl_virginia_users.sql	Updated sequence range, function, is_eu column
ddl_eu_users.sql	Updated Ireland sequence, function, SID-EU prefix, Stockholm added at end
test_virginia_generate_sid.sql	Unchanged — still valid
test_ireland_generate_sid.sql	Unchanged — still valid
test_stockholm_generate_sid.sql	Unchanged — still valid

---

Closes #238