Add continuation support for gz import

[addshore]

Per #6 this one is also needed.

I am currently trying to import all wikibase entities and I am on my third attempt of the import.

[JeroenDeDauw]

Any errors? Why did you have to restart?

[addshore]

The first time was due to having to stop the import and restart it (no errors here) The second time the process stopped (althought I dont know why) as it was running in a screen that autoclosed facepalm

[JeroenDeDauw]

Right. It's quite easy to add an offset parameter that will have the persistence work skipped, though deserialization of entities will still happen. I don't know how the cost of these compares, so am not sure it would make much of a difference.

https://github.com/JeroenDeDauw/Replicator/blob/125b2a5ed3b749215b3dccd76deb37ac098ced22/src/Cli/Command/GzJsonImportCommand.php#L74 (only need changes in this one file)

[addshore]

Well, I imagine this would still speed up getting to the point you left off (no dB transactions!) Are entities not deserialized one at a time? Meaning you could count before deserializing?

[JeroenDeDauw]

They are deserialized one at a time, though if you use this Entity object level iterator, that is hidden from you. You could instead use the string level one and do the deserialization yourself.

This actually suggests it might be good to not hide the fact that the higher level iterators are based on lower level ones in https://github.com/JeroenDeDauw/JsonDumpReader/blob/master/src/JsonDumpFactory.php#L105. Could add two new factory functions there. See https://github.com/JeroenDeDauw/JsonDumpReader/issues/3

[JeroenDeDauw]

The gz stuff actually supports file seek, unlike bz2. Now supported by JsonDumpReader: https://github.com/JeroenDeDauw/JsonDumpReader/commit/9e57c3c0ac5896d4db7365c074c9ae7048236b41

[JeroenDeDauw]

920f03be59c2d15e7342d731409f5c5d8ca2da87