Make concurrency configurable at runtime

[metalmatze]

If we decide to make concurrency configurable at runtime (which we might want to do based on load), I think this will create an unnecessary synchronizer if we have concurrency == 1 for any reason. What do you think of creating a struct with more planning state to make more informed decisions of what to plan? For example, store some sort of notion of whether we have to plan a synchronizer. Maybe it's fine to also check len(phyPlans), but it seems to me like this state complexity will start to balloon. This struct I suggest might also be useful for more planning information (e.g. ordering column guarantees)

Originally posted by @asubiotto in https://github.com/polarsignals/frostdb/pull/202#discussion_r976283462

[github-actions[bot]]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[asubiotto]

I think @thorfour fixed this recently, right?

[thorfour]

I tried, but it ended up being more of a can of worms since our aggregations don't work at concurrency 1. And in testing in prod I ran into additional issues not caught by the unit tests. Needs revisiting.

https://github.com/polarsignals/frostdb/pull/657 for future reference

[github-actions[bot]]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[thorfour]

Removing stale because this is still currently broken.

[github-actions[bot]]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[thorfour]

Closing this in favor of #769