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:

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

To do in follow up PRs

• Update contributing guide #194
• Investigate alternatives for pivotting dataframes #195
• Convert between state- and event-based data representations #196
• Test joining all sample datasets #197
• Fix decode_time error in CI #198
• Simplify licensing #199
• Add notebooks on modelling #200
• Daymet - extract records #201
• Phenocam - what info is relevant? #202
• Simplify installation (instructions) #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.