* core-load-paths.el: Support spacemacs-cache-directory from env variable

[sunlin7]

Fix the #12127 by supporting the spacemacs-cache-directory from environment variable SPACEMACSCACHE or XDG_CACHE_HOME.

NOTE:
This PR can be the first step to start supporting the XDG_* directories to fix #3589.

[fnussbaum]

I think that XDG_CACHE_HOME/spacemacs/ should not be created and used as spacemacs-cache-directory by default, as long as the latter is still used to store data that should persist (and thus should rather be in XDG_{DATA/CONFIG/STATE}_HOME).

[sunlin7]

@fnussbaum Thanks for comment, I had change the code to detect the XDG_CACHE_HOME/spacemacs/ exists or not, please help review again. CC @bcc32

[bcc32]

I think @fnussbaum's comment extends beyond the changes you made. If I understand correctly, we should not use $XDG_CACHE_HOME/spacemacs even if it exists already while persistent state data is still stored using spacemacs-cache-directory. So, we'd have to first audit the places where spacemacs-cache-directory and create a separate variable, perhaps spacemacs-state-directory, for things that should be persisted, before making any of the changes in this PR. (Please feel free to correct me if I misunderstood your comment).

[fnussbaum]

This is a good point, it is indeed better to first clean up the use of the cache directory, thanks @bcc32. (I can see how my comment might have been slightly ambiguous.)

@sunlin7 Thanks for taking the initiative with #16759! I'll need some more time to review it, as I would like to make sure that we do this correctly for all kinds of files, and document it well, such that users that would like to use the new locations can easily migrate, and only have to do it once. It might also make sense to adopt the no-littering package, but I am not so sure about it yet.

[sunlin7]

@fnussbaum Thank you for looking into this, look forward the progress : )