Restore support for commonly used UTF-8 collations

Background

Parent work item: [Remove some charsets](.

To reduce binary size, the parent work item streamlined charset/collation support. Currently, the available collations for utf8mb4 are too few. It is necessary to evaluate and restore the UTF-8 collations commonly used in business scenarios.

Research Conclusion

• utf8mb4_unicode_ci: Common in MySQL 5.x/compatible migration scenarios. Should be prioritized for support.
• utf8mb4_0900_ai_ci: The default collation for MySQL 8.0. Should be prioritized for support in MySQL 8.0 compatible migration scenarios.
• utf8mb4_unicode_520_ci: Used in some legacy systems. Decision on support should be based on compatibility benefits versus binary size increase.
• Further evaluation is possible for utf8mb4_0900_as_cs and utf8mb4_0900_bin. Do not restore language-specific collations in bulk for now.

Implementation Requirements

• Implement the real comparison semantics according to the corresponding Unicode Collation Algorithm version. Must not be simple aliases to utf8mb4_general_ci.
• Correctly handle case sensitivity, accents, and PAD SPACE / NO PAD semantics.
• Ensure compatibility for SHOW COLLATION, DDL, schema dump/restore, and client metadata.

Acceptance Criteria

• Add comprehensive tests covering comparison, sorting, GROUP BY, DISTINCT, unique keys, etc.
• Add compatibility tests for table creation, table alteration, and schema dump/restore.
• Provide the binary size increment introduced by each collation to support future trade-off decisions.

No source files, tests, or entry points are named. Start by locating the existing utf8mb4 collation implementation and compatibility handling, then review how collations are exposed and tested; done means real Unicode comparison behavior, DDL and metadata compatibility, comprehensive comparison and schema tests, and reported binary-size impact for each collation.