[WIP] Add Nextflow example

[jluethi]

Closes #38

It's still work in progress to generalize a bit further. But example 01 runs successfully with this workflow and it already uses the parameters from our parameter json files in the examples.
I'll update it a bit more next week before merging :)

Major parts to explore:

• Currently, the components passed to each task are hard-coded for this example
• Currently, everything is tested for 1 well only
• Currently, nextflow & the tasks run in the same environment and all tasks run in the same conda environment. This can be generalized further
• Currently, every task is defined as its own process, but the processes are very similar. This could potentially be generalized (but maybe that's also what we'd want to have, makes for an easy overview of the Nextflow workflow)
• I haven't figured out where the task logs go so far.

[jluethi]

Quick update on the Nextflow tests: I now managed to get the nextflow example to a state where example 01 can be run successfully through Nextflow with minimal hard-coding: Now, there are a few parameters that need to be set, but nothing hard-coded in the individual processes (each process is 1 task). I wrote some helper functions to parse metadata and components that are run in the processes as well, as part of the script (a bit of pre- and postprocessing).

This makes the processes more task specific now, they don't all have the same content (even besides the different task names being called). Some processes generate metadata, others don't actually add useful updates to them.

I added 2 version now:
run_nextflow.nf updates the metadata after each task. In that way, each task is depending on the prior task and will only execute when that one is finished.
run_nextflow_better_metadata_handling.nf: Only update the metadata in the tasks that run on the plate level. But by default, processes then don't wait for prior processes to finish, so I added dummy outputs that are passed to the next process as an input (clearly not the intended nextflow behavior, couldn't get other things to work so far).

Next steps:
Parallelize over multiple wells: The component logic should help and prepare for this

Hurdles:

1. How do I make some processes wait (e.g. copy-ome-zarr should wait for all yoko-to-ome-zarr processes to finish)?
2. Handling metadata: Would approach 1 still work? Every process write a consolidated metadata, passes it on to the next process? I don't understand nextflow enough to judge this yet.