feat: support MySQL ASCII and UNICODE character set shorthands

[minleejae]

MySQL accepts CAST('ab' AS CHAR(10) ASCII) and JSON_VALUE(j, '$.v' RETURNING CHAR(10) UNICODE), including generated-column expressions. These currently fail in ColDataType. Recognize the shorthand attributes when Dialect.MYSQL is selected, without reserving new keywords.

Refactor character-set spelling into ColDataType.CharacterSetSyntax (CHARACTER_SET, CHARSET, ASCII, UNICODE), retaining the existing isUseCharsetKeyword()/setUseCharsetKeyword() API. getCharacterSet() exposes latin1 for ASCII and ucs2 for UNICODE, while rendering preserves the shorthand. Updating the character set switches a shorthand to an explicit clause; equality and hashing include the spelling.

Example:

ColDataType type = new ColDataType("CHAR").addArgumentsStringList("10")
        .withCharacterSetSyntax(ColDataType.CharacterSetSyntax.ASCII);
// type.getCharacterSet() is "latin1"; type.toString() is "CHAR (10) ASCII"
type.setCharacterSet("utf8mb4"); // renders CHAR (10) CHARACTER SET utf8mb4

Validation: ./gradlew check passed. Tests cover CAST, JSON_VALUE, generated and ordinary columns, construction and edits, equality, dialect boundaries, legacy CHARSET methods, both renderers and reparsing. Ten reproductions were executed on MySQL 8.4.11 before and after rendering, with identical results.

Reference: MySQL character-set shorthand in CAST.

Combined validation with #2682, #2683, #2684, #2685 and #2686: all five changes merge without conflicts; the integrated ./gradlew check passes (8,190 tests, 25 skipped). All 254 MySQL 8.4.11 fixtures parse, render and reparse, and execute with results identical to the original SQL.