(DO NOT MERGE) dbSta: Add delay calculator runtime regressions

[jhkim-pii]

Adds Tcl regressions comparing dmp_ceff_elmore and dmp_ceff_lambert_w with full find_requireds timing updates

• In a few -threads 64 tests, Lambert W dcalc is slower.

STA full timing update runtime

• T1: -threads 1
• T64: -threads t64
• Default: dmp_ceff_elmore delay calculator
• Lambert: dmp_ceff_lambert_w delay calculator
• Runtime unit: ms

Runtime measurement

set_delay_calculator $model
set start_us [clock microseconds]
sta::find_requireds
set elapsed_us [expr {[clock microseconds] - $start_us}]

Regression design	T1 Default	T1 Lambert	T1 delta	T64 Default	T64 Lambert	T64 delta	Default scaling (T1→T64)	Lambert scaling (T1→T64)
asap7_aes	996.0	657.8	-34.0%	309.4	367.7	+18.8%	3.2x	1.8x
asap7_gcd	18.1	11.4	-36.7%	18.1	12.8	-29.4%	1.0x	0.9x
asap7_riscv	345.7	350.0	+1.2%	170.6	186.1	+9.1%	2.0x	1.9x
example1	0.1	0.1	-21.1%	0.1	0.1	-11.6%	0.9x	0.8x
gf180_jpeg	1950.3	1997.1	+2.4%	669.0	791.1	+18.2%	2.9x	2.5x
mock_array	485.4	435.9	-10.2%	359.4	369.2	+2.7%	1.4x	1.2x
nangate45_dynamic_node	338.2	346.0	+2.3%	189.9	210.0	+10.6%	1.8x	1.6x
nangate45_gcd	13.8	9.7	-29.8%	16.7	12.6	-24.2%	0.8x	0.8x
nangate45_ibex	785.8	799.2	+1.7%	405.1	442.6	+9.2%	1.9x	1.8x

How to run asap7/aes with 64 threads

bazel build //:openroad
cd src/dbSta/test
../../../bazel-bin/openroad -threads 64 -no_splash -no_init -exit dcalc_full_update_runtime_asap7_aes.tcl

Example output

...
average  model=dmp_ceff_elmore      runtime_us=    322024 wns=      -77.106266 tns=    -7171.530065
average  model=dmp_ceff_lambert_w   runtime_us=    363738 wns=      -11.645074 tns=      -74.446227
pass

[gemini-code-assist]

Code Review

This pull request introduces a suite of integration tests to compare the full timing-update runtimes of the Elmore and Lambert-W delay calculators across multiple designs, including ASAP7, Nangate45, and GF180. It adds several test scripts, helper utilities, and updates Bazel and CMake configurations to support these pass/fail tests. The review feedback suggests simplifying conditional expressions in the Bazel build file to direct booleans for better readability, and replacing lsearch with remove_from_collection in the Tcl test script to ensure robust handling of OpenSTA collections.