refactor(core-api): move PluginRegistry to core package

[petermetz]

Commit to be reviewed:

commit 6c88f647baf6f5f70e47d2634697d47498e8bed4 (refactor/core-api/move-pluginregistry-to-core-package-380)
Author: Peter Somogyvari <peter.somogyvari@accenture.com>
Date:   Tue Dec 1 17:04:16 2020 -0800

    refactor(core-api): move PluginRegistry to core package
    
    Fixes #380
    
    Signed-off-by: Peter Somogyvari <peter.somogyvari@accenture.com>

[jonathan-m-hamilton]

LGTM

[takeutak]

The design of PluginRegistry depends to the architecture of Cactus, and this is now under the discussion in PR331,368. Therefore, please allow me to review this PR after PR331,368.

[petermetz]

The design of PluginRegistry depends to the architecture of Cactus, and this is now under the discussion in PR331,368. Therefore, please allow me to review this PR after PR331,368.

@takeutak I think we could optimize our review pipeline/development processes in general to be a little more efficient if we didn't make code changes dependent on documentation changes happening first.
With the risk of crossing into philosophy from software engineering: This goes back to my belief that as much as design is important, we are building something that [1] we have never done before so it may be preferable to iterate fast and be willing to refactor often (while we still can, prior to the 1.0 launch) instead of attempting to get it right with up-front design that locks in the implementation. What do you think?

[1]: Production grade, enterprise ready, generic, flexible, pluggable blockchain integration framework with a healthy open source community around it.

[takeutak]

LGTM