Fix Verilator Warnings in bsg_mem and bsg_misc

[infinitymdm]

Non-functional changes in bsg_mem and bsg_misc to fix Verilator lint warnings and 1 syntax error.

• bsg_mem/bsg_mem_multiport_latch_write_banked_bypassing.sv: Fix syntax error due to mismatched end. Looks like it should have been commented out with the rest of the block above it.
• bsg_misc/bsg_mul.sv: Fix incorrect port name. Fixes Verilator PINMISSING warning.
• bsg_mem/bsg_mem_2rw_sync.sv: Fix misleading indentation of assert ... else blocks. Fixes Verilator MISINDENT warnings.
• bsg_mem/bsg_mem_1r1w.sv: Cast parameter to appropriate length before comparison with logic. Fixes Verilator WIDTHEXPAND warning. Also disable verilator UNSIGNED warning on this comparison.
• bsg_misc/bsg_counter_overflow_en.sv: Cast parameter to appropriate length before comparison with logic. Fixes Verilator WIDTHEXPAND warning.

[taylor-bsg]

The Verilator warnings are an interesting case. I am not sure about casting the values to ints since they are 32-bit signed.

[taylor-bsg]

ChatGPT:

2. Type Conversion and Overflow Logic:
   • The line assign overflow_o = (int'(count_o) == max_val_p); uses int'() type casting, which is likely not necessary and could be replaced with ptr_width_lp'() to match the width of count_o:

     assign overflow_o = (count_o == ptr_width_lp'(max_val_p));

   • This avoids unnecessary type conversions and potential mismatches.

[infinitymdm]

I'm thinking the clearest way to do this will be to apply the BSG_WIDTH macro for casting in both files where there's an affected comparison. That seems clearer to me than using ptr_width_lp, at least (though I admit it's a little less efficient for the preprocessor), and provides a way to do this in both files with the WIDTHEXPAND warning. Expect an updated PR shortly.

EDIT: using BSG_WIDTH generated a whole new bevy of warnings, so we're falling back to what ChatGPT suggested (though I still think this is less clear)

[infinitymdm]

The latest changes are in line with your suggestions, but I worry a bit about the safety of shrinking els_p rather than expanding w_addr_i in bsg_mem_1r1w. Especially with the new UNSIGNED warning.