Great questions here! You're going about this the right way.
1) It's approriate when you need it. If you don't see the answer, you have to take time. That's it. Ideally you don't need it for exhibits around 50% of the time.
2) As long as you need. Aim for 10-15 seconds, but, again, if you don't have insights, you can't force it.
3) When you need it. Ideally only when you hit an exhibit, and once in a "general" situation where you're stuck.
4) No you cannot. If the CEO is calling, or walking in, or bumps into you, you cannot take time. This is rare in an interview, but it's good to practice this.
Now, Instead of spending all this energy wondering "am I fast enough, what counts, what are my limits, can I take x seconds", I HIGHLY recommend you focus on just getting better. Here's how:
1) Read the title - and understand it
2) Read the legends - and understand them
3) Remind yourself of the objective / hypothesis in the case, to see where this might fit
4) Find the differences - where does the line graph plummet or spike? Which column is a lot smaller or bigger than the others? Where does change occur? The differences are what matter
5) Talk outloud while interpreting - first, it helps you think and process your thoughts, second, it lets the interviewer provide guidance and course correct if needed.
Best Rote Practice
Rocket Blocks: https://www.rocketblocks.me/
Best Practice Strategy
1) Read the Economist (especially the daily graph and Financial Time frequently
2) Ask case partners to focus particularly on your chart-reading skills (i.e. by providing you with cases with many charts) - Bain and Deloitte cases tend to be chart heavy
Hire a Coach - pick one that suits you best