Move out of the `runPipe`/`runAnalysis` functions the logic that
extracts the container from the vector type. This logic is now
delegated to `containerVectorToTuple` which converts the vector type to
a tuple of references. This allows python's `run{Pipe,Analysis}` to
safely call `nanobind::gil_scoped_release` which releases the GIL for
the entire duration of the `runPipe`/`runAnalysis` execution.
Rework the `AnalysisRunner` and `PipeRunner` structs into single
functions, these do not need to have a template parameter specifying how
to extract containers as that logic has been delegated to a newly
introduced trait called `ExtractContainerFromList`.
Add the needed infrastructures that allow ephemeral classes (`PipeRun`s)
to be wrapped as a regular pypeline `Pipe` class. These automatically
have their dependencies tracked and propagated to the pypeline.
Giacomo Vercesi