grt: replace assert with logger error in CUGR maze route

[Divinesoumyadip]

Fixes #9869

Replaces raw assert statements in src/grt/src/cugr/src/CUGR.cpp with proper logger_->error() calls following OpenROAD conventions. assert(tree != nullptr) is replaced with logger error GRT 610 and assert(constants_.min_routing_layer + 1 < grid_graph_->getNumLayers()) is replaced with logger error GRT 611. Removed unused #include <cassert>. Raw asserts crash silently in release builds while logger_->error() provides proper error reporting consistent with the rest of the codebase.

[gemini-code-assist]

Code Review

This pull request replaces assert calls with logger_->error for better error reporting, which is a good improvement. I've found one potential issue in mazeRoute where using logger_->error could lead to an inconsistent program state. I've suggested using logger_->warn and continuing to the next net to make the router more robust. The other changes look good.

[gemini-code-assist]

Using logger_->error will throw an exception and terminate the mazeRoute function. This will prevent subsequent nets in net_indices from being routed. However, grid_graph_->removeTreeUsage was already called for all these nets in a separate loop before this one (lines 215-217). This will leave the grid graph usage data in an inconsistent state if the exception is caught and the program continues.

To make the router more robust, consider logging a warning and skipping only the failed net. This allows the process to continue for other nets and keeps the program state consistent.

    if (tree == nullptr) {
      logger_->warn(GRT, 610, "Failed to generate Steiner tree for net {}. Skipping net.", net->getName());
      continue;
    }

[Divinesoumyadip]

Fixed

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[Divinesoumyadip]

@eder-matheus Kindly review it .Thnaks in advance.

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"