Use conda-lock

[sverhoeven]

Fixes #164

Also

• upgrades vagrant base box to ubuntu 22.04 which is same as SRC catalog item

To test

1. Spin up a machine either in SRC or with vagrant
2. Check that landing page looks ok and links work
3. Login to JupyterHub
4. Run https://github.com/eWaterCycle/ewatercycle/blob/main/README.md#usage example

[sverhoeven]

2 conda-lock files (one from ewatercycle repo and one for jupyter) can not be be merged. So we first install the conda-lock from ewatercycle repo and then use a conda env file with pinned versions to install jupyter.

[sverhoeven]

To do second vagrant provision I had to fix vagrant see https://github.com/hashicorp/vagrant/pull/13493

[sverhoeven]

roles/ewatercycle/templates/environment.yml.j2 now has direct dependencies pinned.

This should be more reproducible, but you will miss security updates and deps of deps can still break things.

[BSchilperoort]

2 conda-lock files (one from ewatercycle repo and one for jupyter) can not be be merged. So we first install the conda-lock from ewatercycle repo and then use a conda env file with pinned versions to install jupyter.

I guess the alternative would be to built a completely new conda-lock file from scratch? Is that possible/viable/desirable?

[BSchilperoort]

Things work now 👍