feat(java): remove the hadoop dependency from java-info

[SemyonSinchenko]

Describe the enhancement requested

At the moment Hadoop is a dependency of the java-info package. But by the design this package was supposed to be a lightweight and (almost) dependency free. We can, of course, shade all the Hadoop deps but it will make our JAR huge. I'm suggesting to remove Hadoop at all and rely instead on java-core DataInputStream / DataOutputStream that are implemented in Hadoop. For the testing purposes and for the Local FS we can rely on the java built-ins.

With such an improvement a functionality for working with arbitrary file system will be very easy to implement

Component(s)

Java

[SemyonSinchenko]

@Thespica cc

[Thespica]

@Thespica cc

I remembered that I imported the hadoop dependency so that we can read and wire file from different places(e.g. file, s3c ...).

I agree that we can remove this dependency currently, and add what we need and exclusive others if we really need part of the hadoop dependency.

[SemyonSinchenko]

I will work on removing Hadoop dependency in a way when its integration back will be as smooth as possible!