Tutorial + API upgrades

[Peter9192]

This PR represents a massive cleanup and restructuring of springtime.

• Goal: Emphasize and demonstrate how springtime helps to harmonize data sources and models
• How: Add tutorial with notebooks to demonstrate which steps are done 'under the hood' during data load in springtime, and how we've adapted existing modelling packages.
• Side effect: make it easier to use springtime through the API

Changes:

• Less or no common/shared attributes on the abstract dataset class
• Distinguish between preprocessing during load of one dataset, and preprocessing on the joined dataframe
• Models no longer part of recipe or springtime API. Only demonstrate in tutorial/examples.
• Separate loaders for bbox/point/points instead of separate classes (?)
• Distinguish between load and raw_load
• Pydantic V2
• Much slimmer docs
• No longer importing R dependencies!
• Simplify CI: no more calling R and servers, only install springtime python code.

Progress so far:

• [x] Intro
• [x] Combining data
• [x] PEP715
• [x] NPN - ~server down~
• [x] PPO
• [x] Phenocam
• [x] EOBS
• [x] Daymet
• [x] MODIS
• [x] Modelling intro
• [x] Clean up & reorganize docs
• [x] Pydantic V2
• [x] Remove modelling from src/springtime and recipes
• [x] (Re)move example recipes
• [x] Decide on fate of API docs (keep it for now)
• [x] Update CI

To do in follow up PRs

• [x] #194
• [ ] #195
• [ ] #196
• [ ] #197
• [ ] #198
• [x] #199
• [x] #200
• [ ] #201
• [ ] #202
• [x] #205

closes #174 closes #158 closes #156 (no longer planned) closes #153 (no longer planned) closes #111 closes #156 closes #188 closes #190 closes #189 closes #187 should close #186 could close #182

[Peter9192]

There's a few remaining issues in the CI, but I'll postpone those to later dedicated issues/PRs. For now, this is good enough and provides a solid basis to build upon. I'm gonna open new issues for the remaining to do's.