In the hl-wasm scenario we also want to understand the differences between time spent in Wasmtime vs in Hyperlight
we should either add trace exits to log when wasmtime execution starts/stops, or we should build a dummy version of the wasm guest that doesn't actually do the function call into wasmtime but just replays a log of parameter/return interactions; either would give us decently realistic view of how much is hyperlight parameter passing, with slightly different accuracy tradeoffs modes