feat(fts): add code tokenizer with configurable child tokenizer

[egolearner]

Summary

Add an opt-in code tokenizer for identifier-aware full-text search. getRequestTime emits getrequesttime, get, request, and time, allowing keyword queries to retrieve camelCase and snake_case names. Remaining text uses a configurable child tokenizer: standard by default or jieba with nested parameters. All output is lowercased without requiring a lowercase filter.

Preserve existing flat AND/OR, BM25, and phrase execution semantics. Phrases operate on the expanded token sequence. Offset and token-count overflow guards log an error and return the tokens generated so far instead of throwing. Changing tokenizer configuration requires rebuilding the FTS index.

Validation

• Built code_tokenizer_test, fts_parser_test, and fts_query_test on macOS.
• Updated tests cover identifier splitting, Unicode boundaries, child configuration, intrinsic lowercasing, offsets, phrase parsing/retrieval, and FTS index lifecycle.
• Current test execution is blocked locally: all three binaries were terminated with SIGKILL before producing test output. Runtime results remain unverified for this revision.
• git diff --check passed. Linux and Windows execution remains pending.