Aim UI takes a lot of time to load the search results when runs have large amount of runs/metrics

[jgiannuzzi]

This is not yet a minimal reproducible example, but here goes:

• 79 runs, 2261 metrics + 112 params per run, 50 values per metrics
• go to Runs explorer
• server replies in less than a second
• UI freezes for multiple seconds before displaying the runs
• scrolling to the right stutters a lot
• memory and cpu usage are quite high for just some tabular data

Tasks:

• [x] improve how the data gets loaded into the app model by calling getMetricsInitialRowData once per model instead of once per metric/param/tag per run — #437
• [ ] create a proof of concept where data gets served as Arrow records by the backend, stream loaded in the frontend via Arrow, and then served via a virtual data grid — #438
• [ ] based on the proof of concept, document more tasks to solve UI slowness here

[dave-gantenbein]

Let's evaluate the current status of the UI with the data load as described above.

[jescalada]

I've tried diagnosing this by making a k6 load script with the new parameters. I'm getting very slow responses from the server when searching for metrics, assuming we have lots (6000+) of different metrics logged with 15+ iterations for each:

Pinging @suprjinx @dsuhinin for any thoughts about this