Resume file source to complete resources cache path change

[LukasPaczos]

Closes #14334.

I've initially made DefaultFileSource::setResourceCachePath synchronous and managed threading on the java side, but now that I think about it, it might be cleaner to use a callback and do the threading on the core side.

[LukasPaczos]

I've made the changes mentioned in the OP and this is ready for a review.

[alexshalamov]

Should there be a way to communicate failure?

Right now, void OfflineDatabase::changePath(const std::string&) can throw exception, should developer be notified when setResourceCachePath fails?

[LukasPaczos]

What's the way for void OfflineDatabase::changePath(const std::string&) to throw? Is that only the assert(!db) and assert(statements.empty()) assertions in OfflineDatabase::initialize()? If that's the case, we are logging a failure to cleanup beforehand, so I don't think we need to send failure callback just before we crash anyway. Same if we fail to open the database.

[tmpsantos]

I think the underlying problem here is that we don't have a proper mechanism for detecting errors when we failed to open the database/

[LukasPaczos]

Is this something that we'd be able to follow up on in a separate PR?

[alexshalamov]

@LukasPaczos almost every line in OfflineDatabase::initialize throws an exception (db->*, file deletion). ResourcesCachePathChangeCallback already provides onError, that might be used on SDK side to notify developer when changePath failed.

This PR did not introduce that issue, so, I think fix for propagating an error on initialize can be done in separate PR. May be worth creating issue + adding TODO note for PathChangeCallback that refers to an issue.

@tmpsantos wdyt?

[LukasPaczos]

Created #14759 and added a TODO.

[LukasPaczos]

Comments addressed.

[alexshalamov]

lgtm from my side, please don't forget to create an issue for possible failure when database initialization fails.