ENH: Make explode work for sets

[dsaxton]

• closes ENH: Make Series.explode work for sets #35614
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[jreback]

impls and tests look fine, the only issue i have is one of ordering. sets are un-ordered, unlike lists and dicst so this could in theory change between runs?

[dsaxton]

impls and tests look fine, the only issue i have is one of ordering. sets are un-ordered, unlike lists and dicst so this could in theory change between runs?

That's true, the order of the output won't be well-defined and could change between runs

[mroeschke]

Maybe for the test we just use a set with 1 element so its deterministic, and in the docs we can highlight for sets order is not guaranteed.

[dsaxton]

Maybe for the test we just use a set with 1 element so its deterministic, and in the docs we can highlight for sets order is not guaranteed.

Good catch on updating docs. For the tests I'm thinking we'd want multiple elements to show that it actually expands the result (I could add a singleton set case as well to show that it becomes a scalar)? To deal with non-deterministic order we can sort the output.

[jreback]

this looks fine, can you merge master and merge on green.