gpl: min tile congestion inflation

[gudeh]

Summary

This PR transforms an existing fixed value into a tunable TCL parameter (routability_min_congestion_for_inflation). It is also modifies the default from 1.01 to 0.95.

When in routability mode, we check each congestion tile, and we only inflate the instances within such tile if it is above the min congestion variable. In other words, this variable is a threshold to apply the inflation to instances or not, depending on the calculated congestion (RUDY) at the tile.

Type of Change

• Refactoring

Impact

Experiments show we require far less routability reverts with a lower min tile congestion inflation value.

With the lower min congestion the reach of inflation goes further away from the center of a congestion hotspot.

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 signed my commits (DCO).

Related Issues

[gemini-code-assist]

Code Review

This pull request introduces a new option -routability_min_congestion_for_inflation to control the minimum congestion threshold for tile inflation, updates the inflation ratio calculation accordingly, and adds detailed logging and statistics for inflation ratios. The review feedback highlights critical issues including potential division-by-zero errors if the threshold is set to 0.0, a potential crash in the new printGCellInflation function due to unsafe vector indexing, a discrepancy between the documented and actual default values for the new option, and a minor C++20 sorting style improvement.

[gemini-code-assist]

Using std::ranges::sort with iterator pairs is less idiomatic in C++20. It is cleaner and more consistent with other parts of the codebase (such as line 1111) to pass the container directly.

Suggested change

	std::ranges::sort(inflation_ratios.begin(), inflation_ratios.end());
	std::ranges::sort(inflation_ratios);

[gudeh]

We do require less gpl iterations during routability mode with this modification. This plot shows the delta versus master.

data source:https://docs.google.com/spreadsheets/d/1HL--4oBzEjAHPVs7d092-1V5ALQ2_9jb4luOAxUW9M8/edit?usp=sharing

from secure-CI:https://jenkins.openroad.tools/job/OpenROAD-flow-scripts-Private/job/secure-gpl-min-congestion-inflation/

We also go from 42 to 46 designs that successfully achieve the target congestion.

[gudeh]

In the past I opted to not create this PR because I noticed an inverse correlation between routing congestion and timing. Although now this seems at least more subtle.