Issue 47 validate qc part 1

[skymandr]

First of probably many parts of #47. Also brings number of mypy errors below 500! 🥳

What works:

• can load a converted project (using https://github.com/Scan-o-Matic/project-converter/pull/4)
• can interact with plates, see different phenotypes, and mark colonies
• can inspect grid
• can run normalize

What doesn't

• [ ] there are tonnes of console errors from mouse events when interacting with the plate (something with d3.event)
• [x] ~the scan.instructions can't be loaded, which causes some bugs~ (fixed by https://github.com/Scan-o-Matic/project-converter/pull/5)
• [x] ~jsonizer not fully integrated~ (fixed
• [ ] though normalization can be run, I'm not sure it does anything reasonable, and interacting with the normalized phenotypes causes a hang)

In addition the following warnings (and probably others) end up in the docker-compose log:

WARNING **Phenotyper** There are gaps in the time series at times: 42.738247944444446 - 43.406780194444444, 69.11481372222222 - 69.78131019444444
WARNING **Phenotyper** Outdated phenotypes format, guessing which were extracted and converting to current format.To faster load in the future and get rid of this message, save current state.
WARNING **Phenotyper** The plates don't agree on what phenotypes were extracted, you should tread carefully or better yet: rerun feature extraction.
WARNING **Phenotyper State** Filter doesn't match number of plates. Rewriting...

[local-minimum]

* [ ]  the `scan.instructions` can't be loaded, which causes some bugs

Why not just adding it in the converter PR, it should be much easier than what you did with for the other serialized state

[skymandr]

* [ ]  the `scan.instructions` can't be loaded, which causes some bugs

Why not just adding it in the converter PR, it should be much easier than what you did with for the other serialized state

I have started on a PR for that, so far without success. I think this PR is messy enough without it.

[skymandr]

Please see updated PR description for notes!