Improve data flow in the `async` package

[Vasco-jofra]

This PR makes these changes:

1. In FlowSummaryPrivate.qll: Added support for anyProperty content set in flow summaries:
   • Please confirm this is the correct way to support this. I needed this to find the async_.map({a: source()}, call_sink) case. Let me know if there's a better way to do it
   • As a note, I was also unable to use Element, which, according to the documentation, should select "an element of an array, iterator, or set object."
2. In AsyncPackage.qll: Improve taint tracking through functions from the async package:
   • Improve tracking of the callback arguments
   • Implemented flow summaries for more robust taint tracking
   • Updated tests

[Vasco-jofra]

Also, let me know if there's a way to avoid having two DataFlow::SummarizedCallable when all I want to do is change the argument number. I found no way to get a reference to the call in the propagatesFlow predicate against which I could call getIteratorCallbackIndex.

[asgerf]

Also, let me know if there's a way to avoid having two DataFlow::SummarizedCallable when all I want to do is change the argument number. I found no way to get a reference to the call in the propagatesFlow predicate against which I could call getIteratorCallbackIndex.

You can merge the two classes by adding a field to the class, and embedding the value of the field in its this binding. See this example.

[asgerf]

As a note, I was also unable to use Element, which, according to the documentation, should select "an element of an array, iterator, or set object."

Thanks, I'll make sure to fix the discrepancy.

[Vasco-jofra]

I fixed what you pointed out in your comments and merged the DataFlow::SummarizedCallable classes into a single one as you suggested.

I don't have time to test now, but in a codebase I was running this on, there was an asyncHelper.each function that called async.each inside. This change was causing asyncHelper.each(array1, cb1); asyncHelper.each(array2, cb2), to incorrectly flow data from array1 to the first parameter of both cb1 and cb2. This caused several false positives. I haven't investigated why this happens, but we should fix it before merging.