When an unknown signal is encountered, total is set to -1 but the
loop continues. Subsequent known signals add their widths to -1,
producing incorrect totals (e.g. -1 + 16 = 15 instead of -1).
This can mask genuine truncation bugs in status word packing.
The golden reference used (adc_val - 128) << 9 which subtracts 65536,
but the Verilog RTL computes {1'b0,adc,9'b0} - {1'b0,8'hFF,9'b0}/2
which subtracts 0xFF00 = 65280. This creates a constant 256-LSB DC
offset between the golden reference and RTL for all 256 ADC values.
The bit-accurate model in fpga_model.py already uses the correct RTL
formula. This aligns golden_reference.py to match.
Verified: all 256 ADC input values now produce zero offset against
fpga_model.py.
The default IVERILOG and VVP paths were hardcoded to macOS Homebrew
locations (/opt/homebrew/bin/iverilog). On Ubuntu CI runners, apt
installs iverilog to /usr/bin/, so the Path.exists() check returns
False and all Tier 2 Verilog cosim tests are silently skipped.
Change defaults to bare command names so the existing which-based
fallback at line 57-58 discovers the binary via PATH on any platform.