guide: "Comparing Experiments" improvements

[jorgeorpinel]

Extracted from #2862 (review) etc.

Rel. https://dvc.org/doc/user-guide/experiment-management/comparing-experiments

General

• Needs detailed copy editing. E.g. missing "the", disjointed paragraphs, unnecessary scrolling blocks, etc. (another e.g.)
• UG: Write a synopsis section to Comparing Experiments #3160 ?
• Review "baseline" concept. Apply to explanations and branch names?
  Per guide: Comparing Experiments #2862 (review)
• Cover exp show first (exp list later).
  Per guide: Comparing Experiments #2862 (comment)

Specific ideas

• Address guide: revisit Exp Sharing #2908 (review) .
• Cover deps/dependencies column. See Dvc exp show deps #3220
• Absorb #list-experiment-names-to-use-in-scripts into #list-experiments-in-the-workspace.
• Consider tabs when commands are used as alternatives in specific example tasks (e.g. exp show vs exp diff).
• Include/mention --only-changed per ref: exp show --only-changed #2966 .
• Avoid terms "git ref" and "attached". (Use "based on" for consistency?)
• --sort/order-by/no-timestamp sections may be too detailed. Hide or just mention + link to ref.
• Merge JSON and CSV sections more somehow (tabs?).
• Shorten JSON/CSV output sample blocks and remove corresponding schema descriptions (comprehensive examples are enough). (See guide: Comparing Experiments #2862 (review) and below.)
• Format CSV samples if possible (even if that's not how they're output exactly)
• Clarify exp diff sections to separate cases: workspace vs. HEAD, rev A vs rev B, etc. (See guide: Comparing Experiments #2862 (review) and below.)
• Distinguish experiments that ran vs queued ones (exo show/list).

[jorgeorpinel]

@iesahin could you prioritize the general points as part of phase 1?

[iesahin]

I have split the doc into two in #3011 @jorgeorpinel. It may be easier this way, PTAL.

[jorgeorpinel]

Let's discuss as agreed @iesahin 🙂

Its too long. It’s hard to make sense w/o additional help /structure.

💡 instead of grouping sections by command, let's consolidate them into major actions e.g. Review existing experiments can cover basic exp show and all of exp list. Feel free to leave some things out from sample blocks (just mention options, etc.)

The other 2 general check boxes should be straightforward to address. I think.