Make "replicate ls" not get larger and hard to read over time

[bfirsh]

This is covered in detail in #284. In short, as you make more and more experiments with more and more params, each experiment in replicate ls will grow taller and taller, making the output unusable.

Some ideas:

• Lean on experiment grouping as a way to fix this #297
• By default only show most recent experiments, then only params changed recently will be displayed
• Intelligently display only relevant params, somehow