Named worker that is statically guaranteed not to outlive its function

[jclark]

It would be useful to be able to have a named worker that is guaranteed not to outlive the function that created it. This is not currently possible: even if the function worker waits for the named worker, the function worker may panic before the wait completes, and in that case, the named worker will outlive the function worker.

Two reasons why this is useful:

• we can improve on Extend isolated concept to named workers #1095: a named worker in an isolated function can safely access mutable local variables and parameters that are in the function but outside the worker if it it can be guaranteed that the worker does not outlive the function (although the worker strands would all run on the same thread)
• it's good thing generally (a basic principle of structured concurrency)

[jclark]

If we were to use named workers to represent dataflow blocks, then it would be a real pain for named workers not to have full access to parameters.

[jclark]

Also note that allowing workers to access mutable local variables (at the cost of the workers running on the same thread) would also be beneficial for dataflow, and would allow a less "pure" but perhaps more familiar expression of dataflow.