fix(pyplacedb): union fixed placement obstacles

[zhaoxueyan1]

Summary

• keep each fixed macro's real terminal at its body rectangle so instance identity, pin offsets, and DEF write-back remain intact
• union fixed-macro halos with placement and special-net blockages, subtract the fixed body union, and decompose only the residual obstacle geometry into pin-less synthetic terminals
• compute total_space_area from the exact terminal-geometry union clipped to the core, avoiding double-counted overlapping halos and blockages

Verification

• test/tools/ecc/test_pyplacedb_macro_status.py: 2 passed in 0.44s
  • verifies located hard-macro classification
  • verifies overlapping halos and a placement blockage produce non-overlapping synthetic rectangles
  • verifies exact available area and that synthetic terminals do not leak into DEF components
• Kunminghu XSTop placement with cell_padding_x=400, iteration=3000, and routability optimization disabled stopped at iteration 1296 with overflow 0.09878085, below stop_overflow=0.1; the previous representation plateaued around 0.1131

Scope Note

The Kunminghu ECC process later exited during the separate post-placement feature_placement_map() iRT EarlyRouter evaluation after very high memory growth. DreamPlace placement had already completed, but ECC currently runs that optional evaluation before save_data(), so the final DEF was not written. That orchestration and memory issue is outside this ecc-tools geometry change.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.