Show performance of threadpool tasks

[archbirdplus]

trafficstars

Preferably split it up as well:

• chunkgen
• lighting
• whatever else

This would help optimizations.

[IntegratedQuantum]

What exactly would this show? Average time spent?

[archbirdplus]

Average time per task would be useful for optimizing the tasks. Total time would help identify cpu vs gpu performance issues.

[archbirdplus]

Here's what I have so far. The µs/task stat is actually quite stable even without any averaging.

[archbirdplus]

Honestly, 1 millisecond per 32^3 is kind of ridiculous.

[IntegratedQuantum]

Yeah, and keep in mind there is two tasks per chunk, one for lighting/meshing and one for generation. I think lighting takes a big chunk of that combined time. That's why it might be more interesting to see the time per task. Maybe we can add a getTaskId to the VTable to handle these?

[archbirdplus]

My plan was just enum TaskClass.

[IntegratedQuantum]

Wouldn't this make adding new tasks more difficult?

[archbirdplus]

Findings: This is the version I am making a PR for.

[IntegratedQuantum]

fixed by #445