Only upgrade jupyterhub in user env when upgrading tljh, ensure pip>=23.1.2 in user env

[consideRatio]

• Closes Proposal to minimize what TLJH installs in the user environment #872 with a compromise: to not upgrade things in the user environment besides whats truly essential but still install misc things by default on the initial setup. This translates to upgrading jupyterhub based on the hub env specification which should make them match to the latest jupyterhub within the version constraints.

Additional changes

• hub env will always get pip upgraded to latest, user env will get it upgraded to latest if pip is below 23.1.2
• extracted inlined dependencies to a requirements-hub-env.txt file
• added lower bounds to requirements-hub-env.txt to help users understand what they at least have installed from version control
• refactored a section to improve readability

[consideRatio]

From live meeting chat with input from Min and Simon

• We should pip install pip>=(current latest) in both hub env and user env.

EDIT: I did pip install --upgrade pip in hub env, but pip install pip>=23.1.2 in user env.

[consideRatio]

@minrk @manics modified the git history of this PR to make it a bit more slimmed and easy to follow after trying some things and going back on them.

I've updated the title and PR description to reflect this.

Note that it was messy to inspect the hub env as we install the hub env after user env, and that we have unit test for ensure_user_environment function that runs without assumptions on a hub env. So, I went for pip install --upgrade <jupyterhub version constraint from hub env requirements file> instead.

[minrk]

Looks sensible to me. I don't love parsing the jupyterhub requirement with such a strict pattern, but I understand it.

If we impose restrictions on e.g. jupyter-server version in the user env, I'm not sure where that would go.