feat(job): complete deferred job-file features

Implements the remaining items from issue #20:

- version is now forward-permissive: any value >= 1 is accepted; a
  newer-than-known version loads best-effort (unknown fields ignored,
  warning printed) instead of hard-failing on "must be 1"
- from_job input reference: `inputs: [{ from_job: <job> }]` resolves to
  that job's single-file output + format and implies a dependency edge;
  combined depends_on + from_job graph gets topological ordering and
  cycle detection
- logfile size-rotation, on by default (5MB, keep 3), overridable per
  job (log_max_size / log_keep) or file-wide via a top-level defaults block
- new commands: split (schema/table subsetting via select:), inspect
  (rule validation -> markdown/json report, fails job on enforced-rule
  errors), diff (compare exactly two schemas, never fails), scripts-exec
  (run SQL script dirs against a live PostgreSQL database)
- atomic single-file output/report writes (temp file + rename)
- symlink-escape hardening in SafeJoin via EvalSymlinks preflight

Updates docs/JOB_FILES.md and examples/jobs/relspec.yml accordingly.

cmd/relspec/split.go

@@ -205,8 +205,52 @@ func runSplit(cmd *cobra.Command, args []string) error {
 	return nil
 }
 
-// filterDatabase filters the database based on provided criteria
+// splitSelection is the schema/table selection for a split, independent of the
+// CLI flag globals so the job runner can build one directly.
+type splitSelection struct {
+	Schemas        []string
+	Tables         []string
+	ExcludeSchemas []string
+	ExcludeTables  []string
+	DatabaseName   string
+}
+
+// summary renders a one-line human description of the selection.
+func (s splitSelection) summary() string {
+	var parts []string
+	if len(s.Schemas) > 0 {
+		parts = append(parts, "schemas="+strings.Join(s.Schemas, ","))
+	}
+	if len(s.Tables) > 0 {
+		parts = append(parts, "tables="+strings.Join(s.Tables, ","))
+	}
+	if len(s.ExcludeSchemas) > 0 {
+		parts = append(parts, "exclude_schemas="+strings.Join(s.ExcludeSchemas, ","))
+	}
+	if len(s.ExcludeTables) > 0 {
+		parts = append(parts, "exclude_tables="+strings.Join(s.ExcludeTables, ","))
+	}
+	if s.DatabaseName != "" {
+		parts = append(parts, "database_name="+s.DatabaseName)
+	}
+	if len(parts) == 0 {
+		return "(all schemas/tables)"
+	}
+	return strings.Join(parts, " ")
+}
+
+// filterDatabase filters the database based on the CLI split flags.
 func filterDatabase(db *models.Database) (*models.Database, error) {
+	return filterDatabaseSelection(db, splitSelection{
+		Schemas:        parseCommaSeparated(splitSchemas),
+		Tables:         parseCommaSeparated(splitTables),
+		ExcludeSchemas: parseCommaSeparated(splitExcludeSchema),
+		ExcludeTables:  parseCommaSeparated(splitExcludeTables),
+	})
+}
+
+// filterDatabaseSelection filters db down to the schemas/tables named by sel.
+func filterDatabaseSelection(db *models.Database, sel splitSelection) (*models.Database, error) {
 	filteredDB := &models.Database{
 		Name:            db.Name,
 		Description:     db.Description,
@@ -220,11 +264,11 @@ func filterDatabase(db *models.Database) (*models.Database, error) {
 		Domains:         db.Domains, // Keep domains for now
 	}
 
-	// Parse filter flags
-	includeSchemas := parseCommaSeparated(splitSchemas)
-	includeTables := parseCommaSeparated(splitTables)
-	excludeSchemas := parseCommaSeparated(splitExcludeSchema)
-	excludeTables := parseCommaSeparated(splitExcludeTables)
+	// Selection criteria
+	includeSchemas := sel.Schemas
+	includeTables := sel.Tables
+	excludeSchemas := sel.ExcludeSchemas
+	excludeTables := sel.ExcludeTables
 
 	// Convert table names to lowercase for case-insensitive matching
 	includeTablesLower := make(map[string]bool)