Drop edatachain file support and logic around it

[shcheklein]

We can export / import to / from parquet or json files.

What is the reason for cp command to support edachain as a format?

It makes code more complicated, requires additional care, etc.

I suggest we drop it, unless we have a strong reason not to.

cc @dmpetrov @ilongin

[dmpetrov]

It would be great to have these files for codification - committing human readable files to Git.

For now we can use .parquet for transferring datasets. But in general, we would need a special extension to distinguish dataset files from the others. However, the extension needs to be renamed to something like .dc and .dcp (binary parquet version).

For now I'd not prioritize this.

What is the reason for cp command to support edachain as a format?

That's not clear - cp should not use this.