chore(gfql): disambiguate reentry compile-time module naming

[lmeyerov]

Summary

Closes #1333 with a low-risk naming cleanup in Cypher reentry internals:

• move compile-time bounded-reentry helper ownership from reentry/runtime.py to reentry/compiletime.py
• keep reentry/runtime.py as a compatibility re-export shim
• switch lowering call sites/comments/tests to the compile-time module naming

No semantic behavior change intended.

Why

execution.py now clearly represents runtime dataframe behavior, while the former runtime.py was compile-time query rewrite/orchestration logic. This rename reduces maintainer confusion.

Validation

• python -m pytest -q graphistry/tests/compute/gfql/test_ir_type_propagation.py::TestSeamWith1303LoweringSplit graphistry/tests/compute/gfql/cypher/test_lowering_s3_split_guard.py
• ./bin/ruff.sh graphistry/compute/gfql/cypher/lowering.py graphistry/compute/gfql/cypher/reentry/runtime.py graphistry/compute/gfql/cypher/reentry/compiletime.py graphistry/compute/gfql/cypher/reentry/execution.py graphistry/tests/compute/gfql/test_ir_type_propagation.py

Notes

• Added Development changelog entry.
• Review-skill artifacts: plans/issue-1333-reentry-module-naming/review/final-report.md (local plan memory).