[WIP] Remote repositories for versioning

[zhiltsov-max]

Summary

Related to #130, #131

Key changes:

• Added integration with Git and DVC for versioning.
• Added support for remote repository for dataset configuration.
• Added support of remote data sources for project datasets (with DVC. Available HTTP, s3, Git, DVC).
• A project now consists of a number of data sources - local or remote ones.
• Removed support of project's own datasets.

CLI changes:

• Modifying operations on project data (transform, filter, export) are being recorded now. They can be reproduced after with build command.
• Updated config file structure. Old projects can be read, but they will be saved with a new version.
• Updated installation: to install Datumaro with Git and DVC support, add [VCS] suffix: pip install <url>[VCS]
• Added a number of versioning commands in CLI: tag, pull, push, checkout, commit
• Added remote CLI context to interact with data remotes of a project
• Added repo CLI context to interact with bound Git repositories of a project

Library changes:

• Project class has been significantly changed, however, most of the code should work with minimal, or no changes.
• Projects without binding to a local disk are considered detached. In this mode a Project can only interact with locally available data (no remotes) - mostly, exactly the way it was working prior changes. No versioning capabilities is available in this mode.

How to test

Checklist

• I submit my changes into the develop branch
• I have added description of my changes into CHANGELOG
• I have updated the documentation accordingly
• I have added tests to cover my changes
• I have linked related issues)

License

• I submit my code changes under the same MIT License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below)

# Copyright (C) 2020 Intel Corporation
#
# SPDX-License-Identifier: MIT

[zhiltsov-max]

@nmanovic, implemented:

datum create

# addition (url format here: https://dvc.org/doc/command-reference/import-url)
# with auto remotes:
datum add path/ -f image_dir
datum add path/to.json -f coco_instances

# with manual remotes:
datum remote add s3://net.loc -n r1
datum source add remote://r1/path/to.xml  -f cvat

datum filter (not checked)/transform # copying variant
datum export
datum build

datum commit

datum source *
datum remote *

[nmanovic]

@zhiltsov-max , should we close the PR?

[zhiltsov-max]

It will be continued after the first one as "remote sources support".

[leeyh20]

What happened to this PR? Versioning sounds like a very good idea for dataset management

[zhiltsov-max]

@leeyh20, it is split into 2 parts - this one with remotes and #238 with local commands.

[JaviFuentes94]

Any update on this?

[zhiltsov-max]

@JaviFuentes94, not yet - currently, we have no resources for this task. We are welcome for ideas and suggestions on this functionality, though. Could you describe your use cases?