feat(mssql): add MSSQL writer for generating DDL from database schema

- Implement MSSQL writer to generate SQL scripts for creating schemas, tables, and constraints.
- Support for identity columns, indexes, and extended properties.
- Add tests for column definitions, table creation, primary keys, foreign keys, and comments.
- Include testing guide and sample schema for integration tests.

test_data/mssql/TESTING.md

@@ -0,0 +1,286 @@
+# MSSQL Reader and Writer Testing Guide
+
+## Prerequisites
+
+- Docker and Docker Compose installed
+- RelSpec binary built (`make build`)
+- jq (optional, for JSON processing)
+
+## Quick Start
+
+### 1. Start SQL Server Express
+
+```bash
+docker-compose up -d mssql
+
+# Wait for container to be healthy
+docker-compose ps
+
+# Monitor startup logs
+docker-compose logs -f mssql
+```
+
+### 2. Verify Database Creation
+
+```bash
+docker exec -it $(docker-compose ps -q mssql) \
+  /opt/mssql-tools/bin/sqlcmd \
+  -S localhost \
+  -U sa \
+  -P 'StrongPassword123!' \
+  -Q "SELECT name FROM sys.databases WHERE name = 'RelSpecTest'"
+```
+
+## Testing Scenarios
+
+### Scenario 1: Read MSSQL Database to JSON
+
+Read the test schema from MSSQL and export to JSON:
+
+```bash
+./build/relspec convert \
+  --from mssql \
+  --from-conn "sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest" \
+  --to json \
+  --to-path test_output.json
+```
+
+Verify output:
+```bash
+jq '.Schemas[0].Tables | length' test_output.json
+jq '.Schemas[0].Tables[0]' test_output.json
+```
+
+### Scenario 2: Read MSSQL Database to DBML
+
+Convert MSSQL schema to DBML format:
+
+```bash
+./build/relspec convert \
+  --from mssql \
+  --from-conn "sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest" \
+  --to dbml \
+  --to-path test_output.dbml
+```
+
+### Scenario 3: Generate SQL Script (No Direct Execution)
+
+Generate SQL script without executing:
+
+```bash
+./build/relspec convert \
+  --from mssql \
+  --from-conn "sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest" \
+  --to mssql \
+  --to-path test_output.sql
+```
+
+Inspect generated SQL:
+```bash
+head -50 test_output.sql
+```
+
+### Scenario 4: Round-Trip Conversion (MSSQL → JSON → MSSQL)
+
+Test bidirectional conversion:
+
+```bash
+# Step 1: MSSQL → JSON
+./build/relspec convert \
+  --from mssql \
+  --from-conn "sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest" \
+  --to json \
+  --to-path backup.json
+
+# Step 2: JSON → MSSQL SQL
+./build/relspec convert \
+  --from json \
+  --from-path backup.json \
+  --to mssql \
+  --to-path restore.sql
+
+# Inspect SQL
+cat restore.sql | head -50
+```
+
+### Scenario 5: Cross-Database Conversion
+
+If you have PostgreSQL running, test conversion:
+
+```bash
+# MSSQL → PostgreSQL SQL
+./build/relspec convert \
+  --from mssql \
+  --from-conn "sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest" \
+  --to pgsql \
+  --to-path mssql_to_pg.sql
+```
+
+### Scenario 6: Test Type Mappings
+
+Create a JSON file with various types and convert to MSSQL:
+
+```json
+{
+  "Name": "TypeTest",
+  "Schemas": [
+    {
+      "Name": "dbo",
+      "Tables": [
+        {
+          "Name": "type_samples",
+          "Columns": {
+            "id": {
+              "Name": "id",
+              "Type": "int",
+              "AutoIncrement": true,
+              "NotNull": true,
+              "Sequence": 1
+            },
+            "big_num": {
+              "Name": "big_num",
+              "Type": "int64",
+              "Sequence": 2
+            },
+            "is_active": {
+              "Name": "is_active",
+              "Type": "bool",
+              "Sequence": 3
+            },
+            "description": {
+              "Name": "description",
+              "Type": "text",
+              "Sequence": 4
+            },
+            "created_at": {
+              "Name": "created_at",
+              "Type": "timestamp",
+              "NotNull": true,
+              "Default": "GETDATE()",
+              "Sequence": 5
+            },
+            "unique_id": {
+              "Name": "unique_id",
+              "Type": "uuid",
+              "Sequence": 6
+            },
+            "metadata": {
+              "Name": "metadata",
+              "Type": "json",
+              "Sequence": 7
+            },
+            "binary_data": {
+              "Name": "binary_data",
+              "Type": "bytea",
+              "Sequence": 8
+            }
+          },
+          "Constraints": {
+            "PK_type_samples_id": {
+              "Name": "PK_type_samples_id",
+              "Type": "PRIMARY_KEY",
+              "Columns": ["id"]
+            }
+          }
+        }
+      ]
+    }
+  ]
+}
+```
+
+Convert to MSSQL:
+```bash
+./build/relspec convert \
+  --from json \
+  --from-path type_test.json \
+  --to mssql \
+  --to-path type_test.sql
+
+cat type_test.sql
+```
+
+## Cleanup
+
+Stop and remove the SQL Server container:
+
+```bash
+docker-compose down
+
+# Clean up test files
+rm -f test_output.* backup.json restore.sql
+```
+
+## Troubleshooting
+
+### Container won't start
+
+Check Docker daemon is running and database logs:
+```bash
+docker-compose logs mssql
+```
+
+### Connection refused errors
+
+Wait for container to be healthy:
+```bash
+docker-compose ps
+# Wait until STATUS shows "healthy"
+
+# Or check manually
+docker exec -it $(docker-compose ps -q mssql) \
+  /opt/mssql-tools/bin/sqlcmd \
+  -S localhost \
+  -U sa \
+  -P 'StrongPassword123!' \
+  -Q "SELECT @@VERSION"
+```
+
+### Test schema not found
+
+Initialize the test schema:
+```bash
+docker exec -i $(docker-compose ps -q mssql) \
+  /opt/mssql-tools/bin/sqlcmd \
+  -S localhost \
+  -U sa \
+  -P 'StrongPassword123!' \
+  < test_data/mssql/test_schema.sql
+```
+
+### Connection string format issues
+
+Use the correct format for connection strings:
+- Default port: 1433
+- Username: `sa`
+- Password: `StrongPassword123!`
+- Database: `RelSpecTest`
+
+Format: `sqlserver://sa:StrongPassword123!@localhost:1433/RelSpecTest`
+
+## Performance Notes
+
+- Initial reader setup may take a few seconds
+- Type mapping queries are cached within a single read operation
+- Direct execution mode is atomic per table/constraint
+- Large schemas (100+ tables) should complete in under 5 seconds
+
+## Unit Test Verification
+
+Run the MSSQL-specific tests:
+
+```bash
+# Type mapping tests
+go test ./pkg/mssql/... -v
+
+# Reader tests
+go test ./pkg/readers/mssql/... -v
+
+# Writer tests
+go test ./pkg/writers/mssql/... -v
+
+# All together
+go test ./pkg/mssql/... ./pkg/readers/mssql/... ./pkg/writers/mssql/... -v
+```
+
+Expected output: All tests should PASS