Add Unicode character support to parser for identifiers

[nlothian]

Problem

The parser fails to parse library/numerics/testutils.pl because it uses Unicode characters in predicate names:

• δ_inverses_t/5

Error:

syntax_error(No terminal matches 'δ' in the current parser context)
,δ_inverses_t/5
 ^

Root Cause

The Lark grammar in vibeprolog/parser.py defines atoms/identifiers using ASCII character patterns only. Greek letters and other Unicode characters are not recognized as valid atom characters.

Solution

Extend the parser's atom pattern to include Unicode letter categories:

1. Update the ATOM token pattern in the Lark grammar to accept Unicode letters
2. Use Unicode character classes in the regex: [\p{L}] (letters) or similar
3. Ensure the pattern still maintains valid Prolog identifier semantics
4. Test against both ASCII and Unicode predicate names

Implementation Details

• Review vibeprolog/parser.py for the ATOM token definition
• Update the regex pattern to include Unicode letters
• Ensure backward compatibility with ASCII-only identifiers
• Handle Unicode in quoted and unquoted atoms consistently

Testing Requirements

1. Create comprehensive test file tests/test_unicode_identifiers.py covering:

   • Greek letters in predicate names (α, β, γ, δ, etc.)
   • Other Unicode letters (é, ñ, ü, etc.)
   • Unicode in compound terms
   • Unicode in quoted atoms
   • Unicode in variable names (should still be uppercase or underscore)
   • Mixed ASCII and Unicode identifiers
   • Ensure proper distinction between atoms and variables

2. Verify library/numerics/testutils.pl loads successfully

3. Ensure no regressions in existing parser tests

4. Test that semantics are unchanged (Unicode atoms still unify correctly, etc.)

Documentation

Update FEATURES.md to document:

• Parser now supports Unicode letters in atom identifiers
• Examples of valid Unicode atom names
• Any restrictions on which Unicode characters are supported