Need async metadata support when generating R versions of component libraries

[rpkyle]

Currently, Dash for Python leverages __init__.py to identify whether a package has resources which should be loaded asynchronously, e.g.

https://github.com/plotly/dash-core-components/blob/0e1449813c620adfd337568cee8cc55292d76c02/dash_core_components_base/__init__.py#L42-L48

In Dash for R, the internal.R script provides much of the same information, but currently lacks metadata related to async resources:

https://github.com/plotly/dash-core-components/blob/master/R/internal.R

The package generator should be modified to insert the requisite details about async resources when appropriate.

@Marc-Andre-Rivet @alexcjohnson

[alexcjohnson]

related comment I made a few days ago: plotly/dash-bio#430 (comment)

Not for this PR, but at some point - like perhaps, whenever we get around to fully generating __init__.py - this all seems ripe for abstracting into the dash core, like

_js_dist = _dash.development.list_resources(
    ['bundle.js'],
    ['async~{}.js'.format(resource) for resource in async_resources],
    external_prefix='https://unpkg.com/dash-bio@{}/dash_bio'.format(__version__)
)

May still need some tweaking to handle edge cases like async: 'eager'|'lazy', but we should at least be able to auto-detect sourcemaps (which may not exist in release versions #910) and set them all dynamic: True regardless of the resource setting.

Feels like the end goal is that the metadata just has a list of async_resources - maybe by extension it also has a list of resources (sync), as well as dynamic_resources, eager_resources, and lazy_resources?