mpl: update halo/channel naming and usage, pin-aware flag

[joaomai]

Summary

Updates MPL's macro halo/channel configuration into two arguments for mpl::rtl_macro_placer:

• halo_width/-halo_height→ -min_channel_size {width [height]} (height defaults to width if unset)
• use_full_halo→ -pin_aware_channels (default has changed)

Creates odb::set_halo, now used to set halos for any instance (with a convenience flag to set to all macros).

MPL is still aware of halos set in ODB (be it from DEF or by user), however with a functional change: pin-aware channels don't apply to channels created by instance halos. This actually fixes a bug, where it was possible to have min_spacing channels on some sides without pins, and 0 on sides with pins (if -halo_width/-halo_height were unset), which also leads to changes in lots of tests that had this behavior.

Type of Change

• Bug fix
• Refactoring
• Documentation update

Impact

MPL can't create nor update halos anymore, being only concerned with halos already set beforehand or with the new min_size_channel argument.

Verification

• I have verified that the local build succeeds (./etc/Build.sh).
• I have run the relevant tests and they pass.
• My code follows the repository's formatting guidelines.
• I have included tests to prevent regressions.
• I have signed my commits (DCO).

Related Issues

#11062

[gemini-code-assist]

Code Review

This pull request refactors the macro halo handling in the MPL and ODB modules, replacing the deprecated halo commands with a new, unified channel-based interface (-min_channel_size and set_halo). It also introduces pin-aware channel trimming. The review identified a critical bug in the Tcl implementation of set_halo where the flag check was incorrectly formatted, and a high-severity issue in ClusteringEngine where the layer direction was incorrectly derived from the first geometry of a pin rather than the current box. These issues have been noted for correction.

[gadfort]

I think you should add some testing for the new set_halo command

[AcKoucher]

Just some nits to fix. I really like these changes!

[AcKoucher]

Code-wise LGTM.

Some problems in the checks though:

//test/orfs/mock-array:MockArray_4x4_base_cts_path_groups_test  FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_final_path_groups_test FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_macro_layout_test     FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_test                  FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_flat_cts_path_groups_test  FAILED TO BUILD
[...]
//src/mpl/test:mpl_readme_msgs_check-py_test                    FAILED in 0.1s
//src/odb/test:odb_readme_msgs_check-py_test                    FAILED in 0.2s

[joaomai]

//test/orfs/mock-array:MockArray_4x4_base_cts_path_groups_test  FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_final_path_groups_test FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_macro_layout_test     FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_base_test                  FAILED TO BUILD
//test/orfs/mock-array:MockArray_4x4_flat_cts_path_groups_test  FAILED TO BUILD

These tests require a bumping the commit from bazel-orfs, which can only be done after the feature itself is merged.

[joaomai]

Paired with #4493 on ORFS.

[maliberty]

//src/odb/test:set_halo-tcl_test failed

[AcKoucher]

//src/mpl/test:mpl_readme_msgs_check-py_test                    FAILED TO BUILD

There's also some IT issue going on.