Behaviour of --reload

[kitsonk]

Currently, the --reload flag does not invalidate the transpiled versions of any local modules. It only causes cached versions of non-local modules to be invalidated, which of course in turn invalidates their local transpiled versions.

While it might be unlikely, there is still a valid scenario where a user will make a change in another module that impacts the types that are available to a local module that is already transpiled, but is not caught by the compiler because it skips transpilation.

It feels to me it is less surprising that --reload would also invalidate any local modules that are transpiled, irrespective of if their source has changed.

[bartlomieju]

@kitsonk is this issue still relevant?

[kitsonk]

I believe it is.

[ry]

We should just combine --reload and --recompile ... it's more browser-like and simpler.

[kitsonk]

That would work for me... Related to #1582