Docs: Explain metadata operations costs and affects better

[itaiad200]

Explain better how diffs, merges and commits work. Explain how metadata operations are linear to the size of the change. Explain common usage patterns like getting a change summary, running multiple commits or merges on a branch, etc. and how it might impact performance. Something like this section of the S3 documentation could be very helpful for users going to production