feat!: create IceFlow nodes based on application DAG

[JKRhb]

This is the follow-up PR to #107, which attempts to perform an overhaul of the IceFlow API using our application DAG. This PR is still heavily WIP and gradually incorporate further changes made to #107.

[JKRhb]

I think this PR should now be ready for an initial review. The changes might have turned a bit more drastic (and in some places even controversial) than I originally intended, but I think this should point into the right direction.

Some of the most important aspects:

• I removed any kind of publish internal and even the internal consumer queue, so data is immediately published and passed to the application on arrival. As data is temporarily stored inside ndn-svs anyway, the latter aspect should not have negative downsides from my understanding. We can of course discuss this aspect, though, and could add the queue back in, if needed.
• Consumers and Producers are now not intended to be used by a user anymore and are only used internally (and are constructed based on the application graph and a node name). The registration of consumer callbacks and pushing data are now performed by providing the respective edge name.
• The IceflowScaler class now accepts an IceFlow object instead of individual producers and consumers
• Besides a consumer callback, users can also register a "prosumer" callback that itself can access an interface for pushing data to downstream consumers.

Looking forward to your feedback – if you like, I can also walk you through the PR on Monday.

[JKRhb]

Applied some further adjustments to the prosumer callback registration, that allows the user to specify which downstream edge they to push to, in order to make to it possible to push data to multiple downstream edges at once.

[JKRhb]

Noticed a few more spots that might require discussion below.

[JKRhb]

I think after the recent commits, this PR should now be ready for (hopefully) a final review :)

[JKRhb]

(Just spend about 15 minutes debugging some newly appearing errors involving segmentation faults – in the end, the fix came down to a6c21c5 🤦 This is always great fun :D)

[JKRhb]

When the pipeline succeeds, I am now going to finally squash and merge this PR :)