Dash for R should support asynchronous component loading

[rpkyle]

As described in plotly/dash#899, Dash for Python will soon support eager and lazy loading of components to enhance performance. Dash for R should do the same, which will require

• introducing additional logic to parse _js_dist and inspect the async field
• modifying Dash for R to store this attribute in a new (backend) field, as a logical or character string (TRUE, "eager", "lazy")
• providing support for an eager_loading flag
• providing support for a _force_eager_loading flag, e.g. https://github.com/plotly/dash-design-kit/blob/ba0b941425903fbd3d463b83480699489e8f2c91/dash_design_kit/__init__.py#L33
• implementing logic to modify component loading based on the flags above
• writing a test to validate the expected behaviour of the new implementation

Additional requirements (repasted from an earlier comment):

• Dash for R should extract _dash_compatibility_version from component's __init__.py
• may be overridden with Dash option _dash_compatibility_version
• warnings will be displayed to inform the user of the features inactivated by _dash_compatibility_version and which component(s) are responsible for the deprecated mode
• Need to distinguish 'lazy' files from 'dynamic' files for eagerness
• Plotly.js gets differential treatment and eagerness will not just change behavior but also the set of files to load

@Marc-Andre-Rivet @alexcjohnson