sweep: cleanup

[oharboe]

• Introduce OTHER_VARIANTS, variants that are not part
  of the sweep, but used for other testing purposes or
  from SWEEP variants.
• BoomTile has pathological resource usage when not flattened
  regardless of MAX_UNGROUP_SIZE(we have done a sweep to drive
  a wooden stake through the heart of the idea that some magic
  value of MAX_UNGROUP_SIZE will 'fix' hiearchical synthesis for
  megaboom), so non-flattened synthesis is not interesting
  to examine in megaboom. Upcoming work in OpenROAD will probably
  restore the usefulness of studying hierarchical synthesis
  results in OpenROAD since eliminating unused logic after
  hierarchical synthesis is known (from naja) to reduce
  instances in megaboom by 50%.

[oharboe]

@jeffng-or @vvbandeira FYI, some sort of intermittent(?) artifact problem.

[2024-11-17T14:00:58.362Z] [230 / 236] Action results/asap7/BoomTile/base/2_floorplan.odb; 1168s remote-cache, linux-sandbox
[2024-11-17T14:00:58.925Z] WARNING: Remote Cache: 9 errors during bulk transfer:
[2024-11-17T14:00:58.925Z] com.google.devtools.build.lib.remote.http.HttpException: 401 Unauthorized
[2024-11-17T14:00:58.925Z] <?xml version='1.0' encoding='UTF-8'?><Error><Code>AuthenticationRequired</Code><Message>Authentication required.</Message></Error>

[oharboe]

@jeffng-or FYI global route still fails, even with flattened synthesis. Because of the (intermittent?) artifact problems, I can't see the global route result.

[oharboe]

@maliberty @jeffng-or FYI, with this change, where megaboom is flattened and macro placement is from hierarchical synthesis, I get very little global route congestion, a few hundred instead of 10s of thousands.

I'll try a sweep of layer adjustments to see what that does tomorrow.