REST API endpoint for performance optimized writeback with arrow format

With more and more TM1 implementations happening in the cloud, we see more data loads using REST API / TM1py instead of Turbo Integrator.

When using TM1py or the REST API, the best performance you can accomplish right now is to be on par with Turbo Integrator when writing data to cubes.

I suggest implementing a REST API endpoint that offers write performance beyond the speed of TI.

For this specific endpoint, I suggest prioritizing performance over all other aspects. It could be the go-to write method for all TM1 implementations where performance is critical, or data load volumes are enormous.
To gain maximum performance:

- The endpoint should be limited to Admin users and skip all security checks (= behave like TI in that regard)

- Spreading commands in cell values can be ignored

- TM1 can expect data in a columnar data format like Arrow or Parquet (perhaps choose a binary over JSON format here)

- TM1 can expect leaf-level data exclusively

- TM1 can potentially limit function to numeric cells if that helps performance (or provide alternative endpoint for string data)

- TM1 can expect elements in the table/data frame to map 1 to 1 to cube dimensionality. (e.g., a 3-dimensional cube of year, version, products expects a data frame of 4 columns: year, version, product, value

- If it doesn't compromise performance, it would be nice if treatment of not-existing-elements could be chosen: ignore (= minor errors) vs. abort the write operation