perf(html): use indexed parent lookup

[KRRT7]

Summary

Replace repeated linear parent lookups during HTML element reconstruction with a precomputed element ID map.

Performance

This changes parent lookup from O(elements × parents) to O(elements + parents), improving scalability for documents with many hierarchical HTML elements.

[cubic-dev-ai]

No issues found across 2 files

<sub>Shadow auto-approve: would auto-approve. This is a bounded performance refactor that replaces a linear scan with a precomputed dictionary lookup, preserving all existing behavior.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 2 files (changes from recent commits).

<sub>Shadow auto-approve: would auto-approve. Optimizes parent lookup using a precomputed index, preserving behavior. Bounded internal refactor with clear performance benefit.

Re-trigger cubic</sub>

[KRRT7]

@cragwolfe

=========================== short test summary info ============================
FAILED test_unstructured/metrics/test_evaluate.py::test_text_extraction_evaluation - OSError: Cannot save file into a non-existent directory: '/home/runner/work/unstructured/unstructured/test_unstructured/metrics/../../example-docs/test_evaluate_files/test_evaluate_results_cct'
====== 1 failed, 3637 passed, 13 skipped, 2 xfailed in 990.08s (0:16:30) =======
make: *** [Makefile:34: test] Error 1
Error: Process completed with exit code 2.

transient issue.