Implemented already but needs adapting to new approach of Indexer compilation

pandera will allow using pydantic models to define dataframe schemas, we should leverage this to generate pandera schemas for all frame schemas.

plum has some very useful features like using generic classes

Go full functional

It should be fairly simple to adapt the existing REST API server to support the grafana datasource api

Would probably be limited to simple indexes but would be nice for e.g. fsspec mappings

The mongo and tinydb backend allow multiple inserts, we should expose this feature to increase speed.
Suggested approach:
Perform pre insert checks for all documents, then insert all documents that passed at once.

For the API server this would probably mean adding an additional endpoint

queries should be completely self contained.
Mongodb interpolating index is the biggest issue but can be done with aggregation pipeline

This would be super useful for frame-like behavior

Should be pretty easy to implement and useful for simple local storage

this would apply the root validators to labels

The current rframe seems to use pydantic > 2.0.0, for example, BaseModel is only available in > 2.0.0

But the requirement in pyproject.toml is > 1.9.0:

As the requirement is not consistent, we should update accordingly. Otherwise, there will be some conflicts for other packages depending on Pydantic:
rframe 0.2.20 requires pydantic<2.0.0,>=1.9.0, but you have pydantic 2.6.4 which is incompatible.

Most ORMs use Schema.objects namespace for query functions, we should probably do the same.
Maybe Schema.data/Schema.datasets/Schema.dsets

Can reference xedocs for examples

Not happy with current naming.
Maybe:

• rename QueryX classes to XSelection classes
• rename datasource --> dset/dataset/storage