Add table-driven lexer tests for keywords and built-in types

## Why

The lexer currently has no unit tests. Keyword and built-in type classification in `front/lexer/src/ident.rs` is a compact, stable surface that is well suited to a first contribution.

## Scope

- Add table-driven tests in the lexer crate for every language keyword and built-in type spelling.
- Verify exact-boundary behavior: for example, `fun` is a keyword while `funny` is an identifier.
- Cover `ptr` and `array`, which intentionally remain identifier tokens for the type parser.
- Verify the token lexeme and line number as well as the token kind.
- Do not change the Wave grammar or add new keywords.

## Likely files

- `front/lexer/src/ident.rs`
- A new test module under `front/lexer/tests/`

## Completion criteria

- [ ] Every keyword and built-in type in `TokenType` has an assertion.
- [ ] Prefix/suffix identifier cases are covered.
- [ ] `cargo test -p lexer --locked` passes.
- [ ] Existing parser behavior is unchanged.

Start with front/lexer/src/ident.rs and inspect TokenType’s keyword and built-in type spellings. Add table-driven tests in a module under front/lexer/tests/, including exact-boundary cases, ptr and array, token lexemes, and line numbers. Run cargo test -p lexer --locked; done means every listed token is asserted and parser behavior remains unchanged.

Written by the indexing model from the issue text.