Allow queueing of multiple cells in IW

[rchiodo]

Fixes #8022
Fixes #8615

[rchiodo]

Moved the IW fetch to outside of addCode so that multiple calls to addcode wouldn't block on it.

[rchiodo]

The promise chain wasn't necessary here anymore because the IW handles queueing itself now.

[IanMatthewHuff]

I know that you had looked at this before, so I don't want to rehash a debate on this. But I'm not keen on something that changes the way the function works this much being "hidden" in a decorator. I like decorators for stuff like logging or error handling. Something like this feels like a big functionality difference which I wouldn't want changed by a decorator. They can also be a pain to debug at times.

I guess if this PR looks good and the chaining cases are all handed then I won't hold things up. But if we run into issue in the future with this decorator hiding real issues or making debugging hard I'd argue against it.

[rchiodo]

Debugging it isn't hard, you just set breakpoints in the decorator. The reason for it is so we don't have duplicate code for chaining a promise. Not sure how else to implement common code for promise chaining though. A wrapper method instead?

Something like:

public doStuffWithPromise() {
   return chained(async () => {
       await doingstuff();
   })
}

Does that grok better? It's essentially the same thing as the decorator but more explicit.

[IanMatthewHuff]

Probably just anti decorator bias on my part. I read decorators like this:

@thisStuffDoesn'tReallyMatter
@youCanReadTheFunctionJustFineWithoutThis
@probablyJustLoggingOrSomethingLikeThat
function doStuff() {}

This feels more like:

@thisIsReallyKindaABigChangeToHowThisFunctionWorks
function doStuff() {}

But there is no reason that a decorator can't do something like you are doing here. So I vote we go for it. I was thinking it might cause issues with the case where there is a python error in the chain (where we return a false and show a message) but it looks like you added that in a later commit.

[codecov-commenter]

Codecov Report

Merging #8642 (fb00a6e) into main (70cb636) will decrease coverage by 0%.
The diff coverage is 74%.

❗ Current head fb00a6e differs from pull request most recent head 04665fb. Consider uploading reports for the commit 04665fb to get more accurate results

@@          Coverage Diff          @@
##            main   #8642   +/-   ##
=====================================
- Coverage     71%     71%   -1%     
=====================================
  Files        381     382    +1     
  Lines      24484   24513   +29     
  Branches    3755    3762    +7     
=====================================
+ Hits       17508   17515    +7     
- Misses      5465    5479   +14     
- Partials    1511    1519    +8     

Impacted Files	Coverage Δ
src/client/datascience/types.ts	100% <ø> (ø)
...atascience/interactive-window/interactiveWindow.ts	64% <69%> (-3%)	⬇️
...ient/datascience/editor-integration/codewatcher.ts	69% <76%> (-1%)	⬇️
src/client/common/utils/decorators.ts	80% <81%> (+<1%)	⬆️
...t/datascience/errors/interactiveCellResultError.ts	100% <100%> (ø)
...ience/variablesView/variableViewMessageListener.ts	77% <0%> (-23%)	⬇️
...ent/common/application/webviewViews/webviewView.ts	67% <0%> (-11%)	⬇️
...ience/kernel-launcher/localKernelSpecFinderBase.ts	74% <0%> (-3%)	⬇️
src/client/datascience/raw-kernel/rawSocket.ts	82% <0%> (-2%)	⬇️
... and 3 more