Stabilize RCX generate_pattern regression log

[naveenvenk17]

Summary

• Stabilize the rcx.generate_pattern.tcl log comparison by suppressing optional RCX debug shape messages in this standalone regression.
• Keep RCX-0437/RCX-0438 suppressed for the remainder of the short script so subsequent RCX-related commands cannot reintroduce platform-dependent log lines.

Closes #842.

Tests

• git diff --check -- src/rcx/test/generate_pattern.tcl
• openroad -no_splash -no_init -exit /tmp/rcx_suppress_check.tcl (verified RCX-0437/RCX-0438 are suppressed)
• /usr/bin/openroad -no_splash -no_init -exit generate_pattern.tcl from a temporary WSL worktree (script runs and emits no RCX-0437/RCX-0438; local packaged OpenROAD v2.0-17598-ga008522d8 generates artifact differences versus current goldens, so this is not a valid full golden comparison)
• Attempted bazel test //src/rcx/test:generate_pattern-tcl_test --test_timeout=1200 --noshow_progress; blocked in this WSL image by a remote-cached OpenROAD binary requiring GLIBC_2.38.
• Attempted local no-remote-cache Bazel rebuild; blocked by the local hermetic clang/sysroot setup failing to find string.h while compiling zstd.

[gemini-code-assist]

Code Review

This pull request modifies the generate_pattern.tcl test script to suppress specific RCX debug messages during the bench_wires command, aiming to prevent platform-dependent log variations. Feedback suggests that the messages should remain suppressed for the duration of the script to ensure stability for all subsequent RCX-related commands.

[gemini-code-assist]

The messages are unsuppressed before the bench_verilog command is executed. If bench_verilog (which is also an RCX command) triggers these platform-dependent debug messages, the log comparison will remain unstable. It is safer to keep these messages suppressed until the end of the script or at least until all RCX-related commands are finished. Alternatively, since this is a standalone test script, you could omit the unsuppress_message calls entirely as the process terminates shortly after.

[naveenvenk17]

Done in 2ad8ffc: the suppression now remains in effect for the rest of this standalone regression, including bench_verilog.

[maliberty]

I am confused, all that is in the .ok is

[INFO ODB-0227] LEF file: sky130hs/sky130hs.tlef, created 13 layers, 25 vias
[INFO RCX-0055] Finished 252 bench measurements for pattern MET_OVER_MET
[INFO RCX-0055] Finished 396 bench measurements for pattern MET_OVER_MET
[INFO RCX-0007] Finished 240 measurements for pattern MET_UNDER_MET
[INFO RCX-0057] Finished 180 bench measurements for pattern MET_UNDER_MET
[INFO RCX-0058] Finished 182 bench measurements for pattern MET_DIAGUNDER_MET
No differences found.
No differences found.

there is no mention of these messages.

[github-actions]

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

[naveenvenk17]

The checked-in .ok is the desired clean log. RCX-0437/RCX-0438 are optional debug shape messages that showed up in the failing environment but should not be part of the stable golden log, so the suppression is there to keep them out on builds/platforms that emit them. The suppression now stays active through the rest of this standalone script, including bench_verilog.

[maliberty]

debug messages don't print in this test so why do they need to be suppressed? What is the "failing environment"?