GitLab

The original design for this was that when the user selected a
composite command, a specialized CommanderBackend object was
created, and CommanderController would delegate to it to allow
the user to enter details.

In practice the impedance was wrong. We need extra information
for the composite command (how to prompt the user), *don't*
need a lot of other things in the CommanderBackend interface,
need to duplicate logic re: command execution and risk lifetime
issues with nested composite commands.

In this change, composite commands are represented by a pair
of prompt text and a lambda with a signature similar to
CommandSource::GetCommands. All other logic is is handled by
the controller. For a worked example, see the dependent CL
https://chromium-review.googlesource.com/c/chromium/src/+/2535677

Bug: 1014639
Change-Id: I07c172584afcf4288a460362099dbd73b412daa7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2536050Reviewed-by: Elly Fong-Jones <ellyjones@chromium.org>
Commit-Queue: Leonard Grey <lgrey@chromium.org>
Cr-Commit-Position: refs/heads/master@{#827374}