Propagation of request information across thread boundaries

[scottgerring]

Is your feature request related to a problem? Please describe.
In #1620 we are looking at implementing parallel processing for batches to reduce clock time spent handling batch requests. As part of this, we have started considering what happens when we rely on thread-local storage to propagate request context. In the batch case this affects immediately tracing, but more broadly, I think it is worth considering how PT might help support multi-threaded processing, or at the very least, avoid complicating it.

I would suggest the next step here is to consider all areas of PT that use thread local, and if that can be avoided. This may very well be v2 "cleanup style" work. Next, what other areas of PT may benefit from supporting multi-threaded processing, and what would the impact of this be?

Describe the solution you'd like

Describe alternatives you've considered

Additional context

[scottgerring]

There's a great comment from @humanzz that outlines some of the issues that happen with thread pooling in general in a Java-Lambda environment.

[jeromevdl]

Somehow duplicate of #1671