Name forms differ across waveform readers: map names in the harness so cross-format answer hashes compare

When benchmarking several waveform readers (FST via libfst or wellen, other formats) on the same content, signal names often don't match one-to-one, so a per-signal answer hash comparison silently fails or a plan aborts with 'unresolved' names: - vectors may be named 'x[31:0]' in one reader and 'x' in another; - some readers move a '[m:l]' suffix into a separate index field, or parse 'a[7:0]' as scope 'a' with var '[7:0]'; - memory/array scopes like 'mem[0:2047]' may be dropped or split differently. Fix: keep the readers untouched and add a name map in the harness (strip or re-attach the range suffix, normalise array scopes) so every arm reads one plan file. Require every plan name to resolve in every arm before any timed cell runs, and compare per-signal value hashes, not just change counts. One name per id code also avoids alias collisions where one handle serves two names.