v0.4.0 Migrate to unified object-oriented symbol model (v0.4.0)

Refactor core & migrate to unified object-oriented symbol model

BREAKING CHANGE: Completely redesigned internal symbol storage from flat dictionaries
to typed object registries (VariableSymbol, LabelSymbol, ScreenSymbol, ImageSymbol).
Cache structure updated; bump engine version to 0.4.0.

Summary of changes:

• Architecture:

  • Created senpy/symbols.py featuring dedicated dataclasses:
    VariableSymbol, LabelSymbol, ScreenSymbol, and ImageSymbol.
  • Refactored FileScope (models.py) and SenpyAnalyzer (analyzer.py),
    eliminating 13+ fragmented flat dictionaries and sets (declared_labels,
    declared_screens, symbol_declarations, symbol_types, defined_constants, etc.)
    in favor of 4 unified registries: variables, labels, screens, and images.
  • Streamlined _rebuild_aggregates() boilerplate by ~60%, drastically improving
    maintainability and reducing cache desynchronization risks.

• Consumers & Integrations:

  • Refactored LSP Server (server.py) methods: completions, definition,
    hover, and document_symbols to directly consume the new symbol registries.
  • Updated Story Flow Analyzer (flow.py) and Mermaid graph export to inspect
    LabelSymbol properties directly.
  • Modernized checkers.py, python_ast.py, wiki.py, stubs.py, and cache.py
    to work natively with object models.

• Reliability & PEP 8:

  • Replaced broad contextlib.suppress(Exception) with specific, narrowed
    exception tuples (SyntaxError, ValueError, TypeError, OSError, pickle.PickleError, zlib.error)
    in signature.py and cli.py.

• Tests:

  • Updated test assertions across tests/test_analyzer.py and tests/test_lsp.py.
  • All 54 test suites passing cleanly (0.90s).