What is the best version of the visualisation code? I talk about it a bit in #67
Some thoughts:
custom pickling might be needed so that we can load a python object to a certain depth without needing to stream the whole file.
have a jupyter kernel running while people look at their experiments and this unpickles and visualises result blobs. This will be a feature we want to add anyway so that people can query their data in python.
How do people customise their visualisations? People will probably want to customise on a per-experiment basis.
What is the best version of the visualisation code? I talk about it a bit in #67
Some thoughts: