Running `dagger functions` in dagger/ | Dagger | Page 1

Currently working on fixing the caching of the load module step, but planning on looking at the uncached time again after. It seems to get non-linearly worse as we split up into more modules, which is why it recently has gotten way slower.

Based on current metrics I suspect it has to do with running way too much in parallel and overloading CPUs, so it's possible we may need to add some dagger-native support for limiting parallelism (the buildkit support for that doesn't work for us for various reasons).

I did an experiment a few weeks back to try to dag-ify more go build steps which would be even better by deduping way more equivalent work happening during module loads, but shelved for now since it's probably a project in and of itself to get robust. Parallelism limiting is a bit more blunt but also easier to re-use to help fix other similar situations

#Running `dagger functions` in dagger/