Improve fluidity of interactions
Optimize several parts of MuSink interactive to keep lags and latencies small.
-
Use only background / compute threads for the kernel operations. -
Make all calls to query_workspace
asynchronous -
Communicate epsilon and rho via channels, not via query_workspace
-
Prevent axis creation (largely done for rows already) -
The biggest issue: check how precompilation / PackageCompiler
can improve first-time-to-plot behavior