[ENH] API validator classes

I think it would be good to have API contracts as first-class citizens in order to have an explicit, extensible reference point for unified APIs.

Idea: having a concept of different API contracts explicitply in aiod (first class citizen)

Examples:

* sklearn classifier
* sktime forecaster
* sktime TSclassifier
* pytorch-tabular tabular torch model (not sklearn but one layer below)

Requirements:

* API contract is a first class citizen following strategy pattern
* taggable and retrievable
* contract C can be paired with AI algorithm asset A to check "A satisfies C"

Tentative usage mock design:

```python
from sklearn.ensemble import RandomForestClassifier

from aiod.contracts import SklearnClassifier

SklearnClassifier.istypeof(RandomForestClassifier)
# True
# "RandomForestClassifier is an sklearn classifier"

SklearnRegressor.istypeof(RandomForestClassifier)
# False

SklearnClassifier.runtests(RandomForestClassifier)
# runs tests
# returns test summary or printout

# there are two ways to invoke:
# prior knowledge, i.e., tested somewhere and available as database information (e.g., lookup through batteries included clasess)
# actually re-running the tests, via runtests
```

Methods that all "API contracts" have:

* `istypeof(cls)` - returns `True` or `False`
* `runtests` - runs tests for contract

We can change names of course.

Use cases:

* use case 1 - for automation, e.g.,
* run CRON checks on whether estimator A still satisfies a given contract
* run type and API checks on custom estimators
* this ensures API conformance of estimators offered by the platform, i.e., users can trust if the API contract is "green" they can plug into their workflows
* use case 2: plugin developers for their own estimators
* single endpoint for API contract checking for all AI
* (as long as the API contract is being maintained)
* "API checking service"