Mindsdb

- evaluations feature: intended to track ML model performance with relevant accuracy functions that intake predictions and actual/true values.
- buxfixes
- documentation updates

New features:
- REST API endpoints for [Databases](https://github.com/mindsdb/mindsdb/pull/5628), and [Views](https://github.com/mindsdb/mindsdb/pull/5748)
- Postgres API updates

New integrations:
- [Reddit](https://github.com/mindsdb/mindsdb/pull/5682)
- [Plaid API](https://github.com/mindsdb/mindsdb/pull/5619) for balance and transaction endpoints
- [InfluxDB](https://github.com/mindsdb/mindsdb/pull/5754)
- [Google Calendar](https://github.com/mindsdb/mindsdb/pull/5753) for Events
- [Confluence](https://github.com/mindsdb/mindsdb/pull/5718)

- revert Lightwood requirement to be compatible with models trained with previous version

- bugfixes

New handlers:
- Ingres
- SurrealDB
- Cloud Spanner
- GitHub
- EdgelessDB
- Binance

New ML-engines:
- Auto-Sklearn
- LangChain

support for Nixtla's HierarchicalForecast package
updates of documentation
various bugfixes

Description

Changes:
- OpenAI
- image endpoint 5115
- OpenAI handler is permanent 5132
- Set training progress in lightwood handler 4950
- Twitter handler:
- Check twitter connection args for RW and RO modes 5164
- Upload image in tweet 5105
- Statsforecast
- Implement describe() method 5113
- automated model selection 5012
- Logging of using of APIs 5129
- DOCS

New handlers:
- Empress Embedded integration 5209
- Altibase 5206

Fixes:
- Fix missing dependency in Docker image 5222
- Prediction-time temperature for OpenAI models 5232
- Twitter handler: Fix calculating tweet length 5201
- Fix neuralforecast requirements file 5188