Storage: node and channel data should be able to be stored in HDF

[liamhuber]

Or some other storage paradigm.

Forgetting for a moment all the bells and whistles of black-/white-listing different data types, or controlling the depth to which data is stored, or nodes that are smart enough to always re-run because their execution time is trivial compared to their file IO time, or storing nodes that only exist in a notebook, or saving and loading partially executed graphs... At a minimum, we need to be able to save a completed node or workflow-that-only-uses-registered-nodes, and re-load it later in its fully executed state.

Something like this should work and give us the right output data:

from pyiron_workflow import Workflow

wf = wf("my_stuff")
wf.foo = Workflow.create...
...
wf.save()

wf2 = wf("my_stuff")

assert(wf2 is not wf)
wf2.outputs

Where saving and loading worked automatically based off of the workflow label and the current working directory.

[liamhuber]

Just waiting for the bots to work on the latest release of h5io now. Then, assuming the tests are passing, #160 is ready to go and closes this.