// remaining parts.
                // E.g., coreR.string.hello
                //   -> string.hello (drop the first segment)
                //   test.pkg.R.string.hello
                //   -> string.hello (take last two segments, where the size is determined by the size of qualified access minus 1)

 * is imported somewhere else and used without directly referencing the object instance
 * itself:
 *
 * ```kt
 * import Foo.bar
 *
 * object Foo { fun String.bar() {} }
 *
 * fun usage() {
 *   "hello".bar() // this call has implicit 'Foo' dispatch receiver
 * }
 * ```
 */
internal val FirResolvedQualifier.isImplicitDispatchReceiver: Boolean
    get() = source?.kind == KtFakeSourceElementKind.ImplicitReceiver

     * ---
     *
     * Why not just always provide null for name references? In such case, the following case would be a problem:
     *
     * ```kt
     * fun usage(action: String.(Int) -> String) {
     *   "hello".<expr>action</expr>(10)
     * }
     * ```
     *
     * The user might want to know the type of the `action` callback. If we always return null for the named references,