Support SQL large-object names through shared column type parsing

[minleejae]

EvoMaster works around CHARACTER LARGE OBJECT in H2 conditions by replacing the type name with VARCHAR before parsing. Its conversion code notes that a global replacement can also change literal contents. The original condition with CAST('hi' AS CHARACTER LARGE OBJECT(2)) currently fails at LARGE.

Add a shared large-object type-name production to ColDataType, preserving the complete name and existing length properties across CAST, CREATE/ALTER column definitions and parseColDataType. It handles CHARACTER/CHAR/NCHAR LARGE OBJECT, NATIONAL CHARACTER LARGE OBJECT and BINARY LARGE OBJECT using contextual tokens and the existing suffix parser. These names are documented in the H2 type reference.

Tests cover the downstream condition, structured type and length fields, expression traversal, round trips, comments/whitespace, unchanged literal contents and nearby existing type names. No source-text preprocessing is needed.

Validation: Java 17 ./gradlew --console=plain --max-workers=2 check, including the complete test suite, JavaCC choice-conflict gate, Checkstyle, PMD, SpotBugs and Spotless.

[manticore-projects]

Thank you for supporting H2!