Injecting `vertex_gecko` generation/completion data into `results/v2`
This MR generates the generation/completion data for vertex_gecko. The data was generated by some bare bones script that needs to be cleaned up before merging. The data after being created was injected into the already merged v2 data: V2 Generation/Completion Data Collection (!29 - merged). This MR merges the updated v2 data.
The collection process for this data was about 2 1/2 hours (a bit more for generation prompts). Also to note, the max number of output tokens available in the out of the box vertex_gecko model is 64 which is half of what the other models are permitted to output. In hindsight, I could have used the production API endpoint for code suggestions as I am pretty sure vertex_gecko is powering that at the moment (@mray2020 could confirm this), but a quick script seemed the fastest.