[ENH] Refactor and improve `NaiveForecaster`

[MatthewMiddlehurst]

Describe the feature or idea you want to propose

We should refactor the DummyForecaster into NaiveForecaster and improve its features. The variants listed in https://otexts.com/fpp2/simple-methods.html would be a good start.

Describe your proposed solution

• [x] Refactor to NaiveForecaster
• [x] Add mean
• [x] Add seasonality
• [ ] Add drift
• [x] Add testing to tests/test_naive.py
• [x] Update documentation
• [ ] Add to API page and ensure tidiness

Describe alternatives you've considered, if relevant

No response

Additional context

No response

[TinaJin0228]

@aeon-actions-bot assign @TinaJin0228

[TonyBagnall]

I dont think we need all these variants, its just for basic benchmarking

• [x] refactor dummy to naive, default to predict last seen value
• [x] add option to predict the series mean
• [x] add option for seasonal naive: predict the last season value

Makradakis finds seasonality in fit, but for now just pass as a parameter. We can add others later

[TonyBagnall]

@MatthewMiddlehurst think we can close this now? Further naive would be a separate issue?

[MatthewMiddlehurst]

Think this is fine as is