Adds bulk, field-projected accessors to the Python facade so agent workloads can enumerate the application set-at-a-time instead of paying the per-entity reconstruction get_all_methods_in_application() does. Implements all four accessors from #180.

Motivation and Context

On the Neo4j backend, get_methods() fans out into an N+1 reconstruction (per module → per class → per callable, recursing) — tens of thousands of serialized Bolt round-trips on a large app. Catalog-style consumers only need {signature, decorators} per callable plus bodies/call-sites for a chosen few, yet pay to rebuild everything they discard. The SDK offered only one-at-a-time, fully-reconstructed reads, so consumers hand-wrote Cypher (leaking graph schema). These accessors give the SDK ownership of the projected query and return shape. See #180 for the full diagnosis.

How Has This Been Tested?

• New offline unit tests (tests/analysis/python/test_python_bulk_accessors.py) build an in-memory PyApplication and exercise _iter_callables + all four accessors on the in-process backend — no analyzer run, no Neo4j. 5 passed.
• The Neo4j test module gains test_bulk_accessors_parity, asserting the Neo4j backend returns identical overviews / bodies / decorated results / call-sites vs the in-process backend (runs when a server is reachable; skipped otherwise).
• Full tests/analysis/python + tests/models/python: 25 passed, 6 skipped (the skips need a live Neo4j).
• black is not enforced on this repo (12 pre-existing files would reformat at line-length 180), so the code matches the surrounding hand-formatted multi-line style rather than black output; line length verified ≤180.

Breaking Changes

None — additive only. Four new methods on the PythonAnalysisBackend ABC, both backends, and the facade; a new PyCallableOverview model. The session-reuse change preserves _run's observable behavior (same return, same exceptions).

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the Codellm-Devkit Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

Additional context

Commits:

1. feat(python): add bulk/projected accessors — PyCallableOverview + get_callables_overview(), get_method_bodies(signatures), get_decorated_callables(markers). Both backends enumerate the same callable set; a callable is a "method" only when a class declares it directly (mirrors PY_HAS_METHOD).
2. perf(python): reuse one Neo4j read session — the orthogonal quick win from Bulk/projection accessors for the Python facade to avoid N+1 reconstruction on the Neo4j backend #180: _run reused a fresh session per query; now one lazily-opened session for the backend's lifetime, dropped on error. Speeds up the existing reconstruction path too.
3. chore: sync uv.lock — the lockfile still pinned codeanalyzer-typescript 0.4.0 while pyproject was at 0.4.3 (left stale by feat(typescript): add tsc_only toggle and surface synthesized anonymous callables #179); regenerated to agree. No runtime deps added.
4. feat(python): add get_callsites_for — batch call-sites keyed by owning signature (PY_HAS_CALLSITE via OPTIONAL MATCH so call-site-less callables still get an empty-list entry, matching in-process).