What

The timeline panel's data density histogram currently abandons rendering when chunks contain too much data, falling back to no visualization to maintain performance (<1.5ms frame cost, relevant code). This creates confusing gaps in the histogram that don't reflect actual data density, especially when recordings have periods without any logging.

Instead of giving up, I propose we uniformly sample x rows from the chunk and use that to create a better estimate.

Example

In this example I stopped logging data for about 10 seconds between 12:04:40 and 12:04:50: